superpoint_features.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "c6cbWT_q77Kr"
      },
      "source": [
        "##### Import and preprocessing"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "PmAb9Yzno7Pm"
      },
      "source": [
        "only superpoint features"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "GGtXqaYZodI_",
        "outputId": "951da4ca-eb7e-4044-a66c-7ab43731cca9"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Cloning into 'SuperPointPretrainedNetwork'...\n",
            "remote: Enumerating objects: 81, done.\u001b[K\n",
            "remote: Counting objects: 100% (15/15), done.\u001b[K\n",
            "remote: Compressing objects: 100% (6/6), done.\u001b[K\n",
            "remote: Total 81 (delta 12), reused 9 (delta 9), pack-reused 66\u001b[K\n",
            "Unpacking objects: 100% (81/81), done.\n"
          ]
        }
      ],
      "source": [
        "! git clone https://github.com/magicleap/SuperPointPretrainedNetwork.git"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "GUjJkH_5o1UF",
        "outputId": "fffcdf8e-53fb-4ca5-d142-35bf033b99be"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "/content/SuperPointPretrainedNetwork\n"
          ]
        }
      ],
      "source": [
        "cd /content/SuperPointPretrainedNetwork"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "PNZzlpVDpAz8"
      },
      "source": [
        "superpoint features + superglue matching"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "GbZT0M9Pgcdk",
        "outputId": "5697d001-6efa-4d16-95d4-1db6662fd559"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Cloning into 'SuperGluePretrainedNetwork'...\n",
            "remote: Enumerating objects: 182, done.\u001b[K\n",
            "remote: Total 182 (delta 0), reused 0 (delta 0), pack-reused 182\u001b[K\n",
            "Receiving objects: 100% (182/182), 118.85 MiB | 28.92 MiB/s, done.\n",
            "Resolving deltas: 100% (51/51), done.\n",
            "Checking out files: 100% (89/89), done.\n"
          ]
        }
      ],
      "source": [
        "!git clone https://github.com/magicleap/SuperGluePretrainedNetwork.git"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "WNwmBC_do4CV",
        "outputId": "ecd24bec-171b-46f4-e4ac-1819c264399b"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "/content/SuperGluePretrainedNetwork\n"
          ]
        }
      ],
      "source": [
        "cd /content/SuperGluePretrainedNetwork"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "CKGUZebmpEok"
      },
      "source": [
        "imports and functions"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {
        "id": "-Aq2uhvn6o1Q"
      },
      "outputs": [],
      "source": [
        "import numpy as np\n",
        "import pandas as pd\n",
        "import seaborn as sns\n",
        "import matplotlib.pyplot as plt\n",
        "from copy import deepcopy\n",
        "from scipy.spatial import KDTree\n",
        "import cv2\n",
        "from sklearn.preprocessing import normalize\n",
        "\n",
        "import os\n",
        "from google.colab import files\n",
        "from google.colab import drive\n",
        "from zipfile import ZipFile"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "UMd89mOU8i7Y",
        "outputId": "1aaf4547-e15f-4bc3-ec2d-02818e93a389"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mounted at /content/gdrive\n"
          ]
        }
      ],
      "source": [
        "drive.mount('/content/gdrive')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "sA48R1m885zT"
      },
      "outputs": [],
      "source": [
        "# truth\n",
        "TRUTH_DATASET = 'ground_truth_color'\n",
        "\n",
        "with ZipFile(f'/content/gdrive/MyDrive/{TRUTH_DATASET}.zip', 'r') as dataset_zip:\n",
        "    dataset_zip.extractall(f'/content/{TRUTH_DATASET}')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "8DtY7Kyj8tl1"
      },
      "outputs": [],
      "source": [
        "# restored results\n",
        "for model_name in ['ARCNN', 'DNCNN3', 'FBCNN']:\n",
        "    DATA = f'results_{model_name}'\n",
        "    with ZipFile(f'/content/gdrive/MyDrive/{DATA}.zip', 'r') as dataset_zip:\n",
        "        dataset_zip.extractall(f'/content/{DATA}')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 3,
      "metadata": {
        "id": "1fPrrrdOPW6I"
      },
      "outputs": [],
      "source": [
        "# superpoint features\n",
        "for model_name in ['ARCNN', 'DNCNN', 'FBCNN']:\n",
        "    DATA = f'superpoint_{model_name}_10000'\n",
        "    with ZipFile(f'/content/gdrive/MyDrive/{DATA}.zip', 'r') as dataset_zip:\n",
        "        dataset_zip.extractall(f'/content/{DATA}')"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "CUrELo0ng3HV"
      },
      "source": [
        "###### functions"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 4,
      "metadata": {
        "id": "ZjFrkw2489m5"
      },
      "outputs": [],
      "source": [
        "def download_folder_in_zip(dir_to_zip, output_filename, delete_dir_after_download=False):\n",
        "    os.system( \"zip -r {} {}\".format(output_filename, dir_to_zip))\n",
        "    if delete_dir_after_download:\n",
        "        os.system( \"rm -r {}\".format( dir_to_zip ) )\n",
        "    files.download(output_filename)\n",
        "\n",
        "def make_path_txt(TRUTH_DATASET, RESTORED_DATASET, num_elements = 100):\n",
        "    truth_list = []\n",
        "    for image in os.listdir(f'/content/{TRUTH_DATASET}'):\n",
        "        truth_list.append(f'/content/{TRUTH_DATASET}/{image}')\n",
        "    truth_list = sorted(truth_list)[:num_elements]\n",
        "\n",
        "    restored_list = []\n",
        "    for image in os.listdir(f'/content/{RESTORED_DATASET}'):\n",
        "        restored_list.append(f'/content/{RESTORED_DATASET}/{image}')\n",
        "    restored_list = sorted(restored_list)[:num_elements]\n",
        "\n",
        "    path_list = []\n",
        "    for i in range(len(truth_list)):\n",
        "        path_list.append('\\n' + truth_list[i] + ' ' + restored_list[i])\n",
        "    \n",
        "    # remove simbol '\\n' from first string\n",
        "    path_list[0] = path_list[0][1::]\n",
        "\n",
        "    # create files if they dont exists .txt\n",
        "    with open(\"/content/pairs.txt\", \"w+\") as textfile:\n",
        "        for line in path_list:\n",
        "            textfile.write(line)\n",
        "\n",
        "# functions for superpoint coordinates matching\n",
        "def get_cross_matches(matches1, matches2):\n",
        "    \"\"\"\n",
        "    calculate cross match features\n",
        "    \"\"\"\n",
        "    cross_matches = 0\n",
        "    for k1, v1 in matches1.items():\n",
        "        if v1 in matches2:\n",
        "            if matches2[v1] == k1:\n",
        "                cross_matches += 1\n",
        "    return cross_matches\n",
        "\n",
        "def get_key(d, value):\n",
        "    for k, v in d.items():\n",
        "        if np.array_equal(v, value):\n",
        "            return k\n",
        "\n",
        "def get_matches_with_scipy(kp_idx1: dict, kp_idx2: dict, radius = 1.0):\n",
        "    \"\"\"\n",
        "    get simple matches for two dict of coordinates of features of images\n",
        "    using KDTree from scipy for faster running\n",
        "    \"\"\"\n",
        "    matches = {}\n",
        "    for k1, v1 in kp_idx1.items():\n",
        "\n",
        "        # finding nearest with KDTree\n",
        "        tree = KDTree(list(kp_idx2.values()))\n",
        "        nearest = tree.query(v1)\n",
        "        # print('coordinates:', v1, list(kp_idx2.values())[nearest[1]], 'dist value:', nearest[0])\n",
        "\n",
        "        if nearest[0] < radius:\n",
        "            # key of the nearest in kp_idx2\n",
        "            k2 = get_key(kp_idx2, list(kp_idx2.values())[nearest[1]])\n",
        "\n",
        "            # save matching\n",
        "            matches[k1] = k2\n",
        "\n",
        "            # delete matching key\n",
        "            del kp_idx2[k2]\n",
        "\n",
        "    return matches\n",
        "\n",
        "def superpoint_coord_match(keypoints1: dict, keypoints2: dict, radius = 1.0):\n",
        "    \"\"\"\n",
        "    cross matching value using coordinates of features caclulated by superpoint\n",
        "    \"\"\"\n",
        "    kp1 = deepcopy(keypoints1)\n",
        "    kp2 = deepcopy(keypoints2)\n",
        "\n",
        "    print('radius value:', radius)\n",
        "    print('number of features for img1:', len(kp1))\n",
        "    print('number of features for img2:', len(kp2))\n",
        "\n",
        "    # simple match\n",
        "    im1_VS_im2_match = get_matches_with_scipy(kp1, kp2, radius = radius)\n",
        "\n",
        "    kp1 = deepcopy(keypoints1)\n",
        "    kp2 = deepcopy(keypoints2)\n",
        "\n",
        "    im2_VS_im1_match = get_matches_with_scipy(kp2, kp1, radius = radius)\n",
        "\n",
        "    print('number of matches 1 VS 2:', len(im1_VS_im2_match))\n",
        "    print('number of matches 2 VS 1:', len(im2_VS_im1_match))\n",
        "\n",
        "    # cross_match\n",
        "    cross_matches = get_cross_matches(im1_VS_im2_match, im2_VS_im1_match)\n",
        "    cross_match_value = cross_matches/len(keypoints2)\n",
        "    print('cross matches:', cross_matches)\n",
        "    print('cross match value:', cross_match_value)\n",
        "\n",
        "    return cross_match_value\n",
        "\n",
        "def superpoint_coord_batch_match(keypoints1, keypoints2, radius = 1):\n",
        "    '''\n",
        "    coordinates based match for package\n",
        "    '''\n",
        "    kp1 = deepcopy(keypoints1)\n",
        "    kp2 = deepcopy(keypoints2)\n",
        "\n",
        "    match_values = []\n",
        "    for i in range(len(kp1)):\n",
        "        print('#'*50)\n",
        "        print('pair', i)\n",
        "        print()\n",
        "        value = superpoint_coord_match(dict(enumerate(kp1[i])), \n",
        "                                       dict(enumerate(kp2[i])), \n",
        "                                       radius = radius)\n",
        "        print()\n",
        "        match_values.append(value)\n",
        "    return match_values\n",
        "\n",
        "def set_coord_des(kp, des):\n",
        "    coord = {}\n",
        "    descript = {}\n",
        "    for i in range(len(kp)):\n",
        "        coord[i] = np.array(kp[i].pt)\n",
        "\n",
        "        descript[i] = des[i]\n",
        "    return coord, descript\n",
        "\n",
        "def get_idx_cross_matches(matches1, matches2):\n",
        "    \"\"\"\n",
        "    calculate cross match features\n",
        "    return matching idx\n",
        "    \"\"\"\n",
        "    cross_matches = {}\n",
        "    for k1, v1 in matches1.items():\n",
        "        if v1 in matches2:\n",
        "            if matches2[v1] == k1:\n",
        "                cross_matches[v1] = k1\n",
        "    return cross_matches\n",
        "\n",
        "def dist_between_matched_descriptors(superpoint_kp_truth, superpoint_kp_arcnn, radius=1):\n",
        "\n",
        "    kp1 = deepcopy(superpoint_kp_truth)\n",
        "    kp2 = deepcopy(superpoint_kp_arcnn)\n",
        "\n",
        "    print('radius value:', radius)\n",
        "    print('number of features:', len(kp1))\n",
        "\n",
        "    im1_VS_im2_match = get_matches_with_scipy(kp1, kp2, radius = radius)\n",
        "\n",
        "    kp1 = deepcopy(superpoint_kp_truth)\n",
        "    kp2 = deepcopy(superpoint_kp_arcnn)\n",
        "\n",
        "    im2_VS_im1_match = get_matches_with_scipy(kp2, kp1, radius = radius)\n",
        "\n",
        "    # cross_match\n",
        "    cross_matches = get_idx_cross_matches(im1_VS_im2_match, im2_VS_im1_match)\n",
        "\n",
        "    print('number of matching features:', len(cross_matches))\n",
        "\n",
        "    return cross_matches"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8ND-0DPLNIJ-"
      },
      "source": [
        "### superglue matching (+superpoint featuring)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gOaXehkm63Ph"
      },
      "source": [
        "###### running"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "NtvyZ6YYFRVe",
        "outputId": "5e4e70d4-d77a-4e36-bdc1-12499e7a1d84"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Namespace(cache=False, eval=False, fast_viz=False, force_cpu=False, input_dir='assets/scannet_sample_images/', input_pairs='/content/pairs.txt', keypoint_threshold=0.005, match_threshold=0.1, max_keypoints=400, max_length=-1, nms_radius=1, opencv_display=False, output_dir='/content/superpoint_ARCNN', resize=[-1], resize_float=False, show_keypoints=False, shuffle=False, sinkhorn_iterations=20, superglue='indoor', viz=True, viz_extension='png')\n",
            "Will not resize images\n",
            "Running inference on device \"cpu\"\n",
            "Loaded SuperPoint model\n",
            "Loaded SuperGlue model (\"indoor\" weights)\n",
            "Looking for data in directory \"assets/scannet_sample_images\"\n",
            "Will write matches to directory \"/content/superpoint_ARCNN\"\n",
            "Will write visualization images to directory \"/content/superpoint_ARCNN\"\n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair     0 of   100] load_image=0.223 matcher=26.163 viz_match=2.131 total=28.517 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair     1 of   100] load_image=0.213 matcher=25.608 viz_match=2.103 total=27.924 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair     2 of   100] load_image=0.205 matcher=25.549 viz_match=2.079 total=27.833 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair     3 of   100] load_image=0.198 matcher=25.176 viz_match=2.111 total=27.485 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair     4 of   100] load_image=0.195 matcher=24.933 viz_match=2.110 total=27.237 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair     5 of   100] load_image=0.199 matcher=25.224 viz_match=2.130 total=27.552 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair     6 of   100] load_image=0.200 matcher=24.978 viz_match=2.027 total=27.205 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair     7 of   100] load_image=0.210 matcher=24.646 viz_match=2.036 total=26.892 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair     8 of   100] load_image=0.206 matcher=24.882 viz_match=2.055 total=27.143 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair     9 of   100] load_image=0.202 matcher=24.662 viz_match=2.088 total=26.952 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    10 of   100] load_image=0.201 matcher=24.626 viz_match=2.088 total=26.915 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    11 of   100] load_image=0.203 matcher=25.014 viz_match=2.046 total=27.263 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    12 of   100] load_image=0.218 matcher=24.892 viz_match=1.936 total=27.046 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    13 of   100] load_image=0.213 matcher=24.642 viz_match=1.870 total=26.724 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    14 of   100] load_image=0.206 matcher=24.491 viz_match=2.058 total=26.755 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    15 of   100] load_image=0.232 matcher=24.527 viz_match=2.095 total=26.854 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    16 of   100] load_image=0.222 matcher=24.519 viz_match=2.090 total=26.831 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    17 of   100] load_image=0.213 matcher=24.590 viz_match=2.008 total=26.811 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    18 of   100] load_image=0.231 matcher=24.822 viz_match=2.020 total=27.073 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    19 of   100] load_image=0.218 matcher=24.962 viz_match=2.038 total=27.218 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    20 of   100] load_image=0.210 matcher=24.659 viz_match=2.082 total=26.952 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    21 of   100] load_image=0.206 matcher=25.070 viz_match=2.089 total=27.364 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    22 of   100] load_image=0.202 matcher=24.871 viz_match=2.052 total=27.126 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    23 of   100] load_image=0.215 matcher=24.800 viz_match=1.950 total=26.965 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    24 of   100] load_image=0.211 matcher=25.018 viz_match=1.874 total=27.103 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    25 of   100] load_image=0.208 matcher=24.760 viz_match=1.922 total=26.890 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    26 of   100] load_image=0.205 matcher=24.615 viz_match=1.987 total=26.806 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    27 of   100] load_image=0.201 matcher=24.886 viz_match=2.011 total=27.098 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    28 of   100] load_image=0.199 matcher=24.746 viz_match=1.946 total=26.892 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    29 of   100] load_image=0.212 matcher=24.494 viz_match=1.980 total=26.687 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    30 of   100] load_image=0.206 matcher=24.806 viz_match=2.012 total=27.024 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    31 of   100] load_image=0.201 matcher=24.659 viz_match=2.063 total=26.923 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    32 of   100] load_image=0.199 matcher=24.555 viz_match=2.064 total=26.818 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    33 of   100] load_image=0.198 matcher=24.921 viz_match=2.028 total=27.146 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    34 of   100] load_image=0.211 matcher=24.724 viz_match=1.922 total=26.857 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    35 of   100] load_image=0.206 matcher=24.503 viz_match=1.856 total=26.565 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    36 of   100] load_image=0.203 matcher=24.933 viz_match=1.906 total=27.042 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    37 of   100] load_image=0.202 matcher=24.644 viz_match=1.981 total=26.827 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    38 of   100] load_image=0.200 matcher=24.458 viz_match=2.002 total=26.659 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    39 of   100] load_image=0.199 matcher=24.775 viz_match=1.926 total=26.900 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    40 of   100] load_image=0.217 matcher=24.396 viz_match=1.952 total=26.565 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    41 of   100] load_image=0.209 matcher=24.246 viz_match=2.021 total=26.477 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    42 of   100] load_image=0.204 matcher=25.501 viz_match=2.039 total=27.745 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    43 of   100] load_image=0.200 matcher=25.075 viz_match=2.065 total=27.339 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    44 of   100] load_image=0.202 matcher=24.780 viz_match=2.024 total=27.006 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    45 of   100] load_image=0.212 matcher=25.073 viz_match=1.916 total=27.201 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    46 of   100] load_image=0.205 matcher=24.690 viz_match=1.849 total=26.744 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    47 of   100] load_image=0.200 matcher=24.390 viz_match=1.897 total=26.487 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    48 of   100] load_image=0.196 matcher=24.396 viz_match=2.124 total=26.717 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    49 of   100] load_image=0.221 matcher=24.381 viz_match=2.104 total=26.705 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    50 of   100] load_image=0.213 matcher=24.439 viz_match=2.017 total=26.669 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    51 of   100] load_image=0.224 matcher=24.375 viz_match=2.038 total=26.637 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    52 of   100] load_image=0.222 matcher=24.792 viz_match=2.065 total=27.078 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    53 of   100] load_image=0.215 matcher=24.733 viz_match=2.070 total=27.018 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    54 of   100] load_image=0.207 matcher=24.658 viz_match=2.084 total=26.949 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    55 of   100] load_image=0.204 matcher=25.042 viz_match=2.044 total=27.289 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    56 of   100] load_image=0.217 matcher=24.928 viz_match=1.949 total=27.094 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    57 of   100] load_image=0.210 matcher=24.720 viz_match=1.875 total=26.805 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    58 of   100] load_image=0.206 matcher=24.866 viz_match=1.924 total=26.996 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    59 of   100] load_image=0.202 matcher=24.627 viz_match=1.986 total=26.815 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    60 of   100] load_image=0.197 matcher=24.451 viz_match=1.999 total=26.647 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    61 of   100] load_image=0.199 matcher=24.756 viz_match=1.934 total=26.890 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    62 of   100] load_image=0.208 matcher=24.454 viz_match=1.962 total=26.624 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    63 of   100] load_image=0.203 matcher=24.309 viz_match=1.998 total=26.510 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    64 of   100] load_image=0.199 matcher=24.700 viz_match=2.385 total=27.284 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    65 of   100] load_image=0.259 matcher=25.134 viz_match=2.302 total=27.695 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    66 of   100] load_image=0.240 matcher=24.757 viz_match=2.191 total=27.188 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    67 of   100] load_image=0.241 matcher=25.051 viz_match=2.036 total=27.327 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    68 of   100] load_image=0.228 matcher=24.707 viz_match=1.922 total=26.857 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    69 of   100] load_image=0.215 matcher=24.410 viz_match=1.973 total=26.599 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    70 of   100] load_image=0.210 matcher=24.375 viz_match=2.161 total=26.746 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    71 of   100] load_image=0.230 matcher=24.358 viz_match=2.125 total=26.713 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    72 of   100] load_image=0.218 matcher=24.295 viz_match=2.015 total=26.529 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    73 of   100] load_image=0.226 matcher=24.167 viz_match=2.030 total=26.422 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    74 of   100] load_image=0.216 matcher=24.475 viz_match=2.059 total=26.750 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    75 of   100] load_image=0.206 matcher=24.279 viz_match=2.056 total=26.541 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    76 of   100] load_image=0.201 matcher=24.150 viz_match=2.074 total=26.424 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    77 of   100] load_image=0.197 matcher=24.587 viz_match=2.037 total=26.820 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    78 of   100] load_image=0.214 matcher=24.548 viz_match=1.930 total=26.691 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    79 of   100] load_image=0.207 matcher=24.358 viz_match=1.858 total=26.423 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    80 of   100] load_image=0.201 matcher=24.717 viz_match=1.918 total=26.836 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    81 of   100] load_image=0.198 matcher=24.458 viz_match=1.975 total=26.630 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    82 of   100] load_image=0.196 matcher=24.275 viz_match=1.993 total=26.464 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    83 of   100] load_image=0.193 matcher=24.596 viz_match=1.927 total=26.715 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    84 of   100] load_image=0.201 matcher=24.402 viz_match=1.862 total=26.465 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    85 of   100] load_image=0.208 matcher=24.418 viz_match=1.825 total=26.450 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    86 of   100] load_image=0.211 matcher=24.858 viz_match=1.915 total=26.985 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    87 of   100] load_image=0.207 matcher=25.519 viz_match=1.951 total=27.678 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    88 of   100] load_image=0.205 matcher=25.541 viz_match=1.967 total=27.713 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    89 of   100] load_image=0.218 matcher=25.776 viz_match=1.887 total=27.881 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    90 of   100] load_image=0.219 matcher=25.319 viz_match=1.946 total=27.484 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    91 of   100] load_image=0.212 matcher=24.868 viz_match=1.992 total=27.072 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    92 of   100] load_image=0.207 matcher=24.660 viz_match=2.085 total=26.952 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    93 of   100] load_image=0.228 matcher=24.748 viz_match=2.082 total=27.058 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    94 of   100] load_image=0.216 matcher=24.610 viz_match=1.983 total=26.808 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    95 of   100] load_image=0.231 matcher=24.295 viz_match=2.004 total=26.530 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    96 of   100] load_image=0.220 matcher=24.576 viz_match=2.075 total=26.872 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    97 of   100] load_image=0.212 matcher=24.441 viz_match=2.103 total=26.757 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    98 of   100] load_image=0.206 matcher=24.397 viz_match=2.121 total=26.724 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    99 of   100] load_image=0.203 matcher=24.684 viz_match=2.072 total=26.959 sec {0.0 FPS} \n"
          ]
        },
        {
          "ename": "ValueError",
          "evalue": "ignored",
          "output_type": "error",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mValueError\u001b[0m                                Traceback (most recent call last)",
            "\u001b[0;32m<timed exec>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/numpy/lib/npyio.py\u001b[0m in \u001b[0;36mload\u001b[0;34m(file, mmap_mode, allow_pickle, fix_imports, encoding)\u001b[0m\n\u001b[1;32m    443\u001b[0m             \u001b[0;31m# Try a pickle\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    444\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0mallow_pickle\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 445\u001b[0;31m                 raise ValueError(\"Cannot load file containing pickled data \"\n\u001b[0m\u001b[1;32m    446\u001b[0m                                  \"when allow_pickle=False\")\n\u001b[1;32m    447\u001b[0m             \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mValueError\u001b[0m: Cannot load file containing pickled data when allow_pickle=False"
          ]
        }
      ],
      "source": [
        "%%time\n",
        "# ARCNN\n",
        "make_path_txt('ground_truth_color', 'results_ARCNN')\n",
        "\n",
        "!python match_pairs.py --input_pairs /content/pairs.txt --output_dir /content/superpoint_ARCNN --nms_radius 1 --resize -1 --max_keypoints 400 --viz --match_threshold 0.1\n",
        "\n",
        "match_values_a = []\n",
        "path = '/content/superpoint_ARCNN'\n",
        "for pair in os.listdir(path):\n",
        "    npz = np.load(path+'/'+pair)\n",
        "    match_values_a.append(np.sum(npz['matches']>-1))\n",
        "print('ARCNN mean value:', np.mean(np.array(match_values_a)))\n",
        "download_folder_in_zip('/content/results_match_ARCNN', 'superpoint_ARCNN.zip')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "_3Gwgj37DO8B",
        "outputId": "8b53af1b-561a-4197-f2f8-939e1341864d"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Namespace(cache=False, eval=False, fast_viz=False, force_cpu=False, input_dir='assets/scannet_sample_images/', input_pairs='/content/pairs.txt', keypoint_threshold=0.005, match_threshold=0.2, max_keypoints=10000, max_length=-1, nms_radius=1, opencv_display=False, output_dir='/content/superpoint_DNCNN', resize=[-1], resize_float=False, show_keypoints=False, shuffle=False, sinkhorn_iterations=20, superglue='indoor', viz=True, viz_extension='png')\n",
            "Will not resize images\n",
            "Running inference on device \"cpu\"\n",
            "Loaded SuperPoint model\n",
            "Loaded SuperGlue model (\"indoor\" weights)\n",
            "Looking for data in directory \"assets/scannet_sample_images\"\n",
            "Will write matches to directory \"/content/superpoint_DNCNN\"\n",
            "Will write visualization images to directory \"/content/superpoint_DNCNN\"\n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "tcmalloc: large alloc 1600004096 bytes == 0x14bb0000 @  0x7f6dcc025b6b 0x7f6dcc045379 0x7f6d546ded57 0x7f6d546ccbc3 0x7f6d7e5d76af 0x7f6d7e5d8020 0x7f6d7e5d8074 0x7f6d7e5d81bf 0x7f6d7f35f82b 0x7f6d7f3c34c2 0x7f6d7e913726 0x7f6d7e8fdf4a 0x7f6d7f36436e 0x7f6d7f3643c0 0x7f6d7f1a3a3c 0x7f6d805be279 0x7f6d805becc3 0x7f6d7f1e78c4 0x7f6d7e8df8ad 0x7f6d7e8e4b4a 0x7f6d7f5a5b6a 0x7f6d7f0e3135 0x7f6da6346ad6 0x593784 0x548c51 0x51566f 0x549e0e 0x593fce 0x548ae9 0x51566f 0x593dd7\n",
            "tcmalloc: large alloc 1600004096 bytes == 0x74192000 @  0x7f6dcc025b6b 0x7f6dcc045379 0x7f6d546ded57 0x7f6d546ccbc3 0x7f6d7e5d76af 0x7f6d7e5d8020 0x7f6d7e5d8074 0x7f6d7e5d81bf 0x7f6d7f35f82b 0x7f6d7f3c7e36 0x7f6d7e61d1cf 0x7f6d7e61eb83 0x7f6d7e61fd70 0x7f6d7f388325 0x7f6d7f388390 0x7f6d7ef6a99c 0x7f6d80468a68 0x7f6d804691e3 0x7f6d7ef9e1a4 0x7f6da6295cd4 0x7f6da6295f37 0x593a14 0x594cd3 0x53124b 0x4d1138 0x5132c4 0x593dd7 0x5118f8 0x549e0e 0x4bcb19 0x59c019\n",
            "tcmalloc: large alloc 1600004096 bytes == 0x14bb0000 @  0x7f6dcc025b6b 0x7f6dcc045379 0x7f6d546ded57 0x7f6d546ccbc3 0x7f6d7e5d76af 0x7f6d7e5d8020 0x7f6d7e5d8074 0x7f6d7e5d81bf 0x7f6d7f35f82b 0x7f6d7f3c0e52 0x7f6d7ea6129a 0x7f6d7f364c65 0x7f6d7f364cc4 0x7f6d7f1a5020 0x7f6d805b11a8 0x7f6d805b1777 0x7f6d7f20aead 0x7f6d7ea626c8 0x7f6d7f5ab143 0x7f6d7f1f635e 0x7f6da628d9fb 0x59b1b0 0x515655 0x549576 0x593fce 0x548ae9 0x5127f1 0x593dd7 0x5118f8 0x549e0e 0x4bcb19\n",
            "tcmalloc: large alloc 1600004096 bytes == 0x12e64000 @  0x7f6dcc025b6b 0x7f6dcc045379 0x7f6d546ded57 0x7f6d546ccbc3 0x7f6d7e5d76af 0x7f6d7e5d8020 0x7f6d7e5d8074 0x7f6d7e5d81bf 0x7f6d7f35f82b 0x7f6d7f3c34c2 0x7f6d7e913726 0x7f6d7e8fdf4a 0x7f6d7f36436e 0x7f6d7f3643c0 0x7f6d7f1a3a3c 0x7f6d805be279 0x7f6d805becc3 0x7f6d7f1e78c4 0x7f6d7e8df8ad 0x7f6d7e8e4b4a 0x7f6d7f5a5b6a 0x7f6d7f0e3135 0x7f6da6346ad6 0x593784 0x548c51 0x51566f 0x549e0e 0x593fce 0x548ae9 0x51566f 0x593dd7\n",
            "Traceback (most recent call last):\n",
            "  File \"match_pairs.py\", line 274, in <module>\n",
            "    pred = matching({'image0': inp0, 'image1': inp1})\n",
            "  File \"/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py\", line 1130, in _call_impl\n",
            "    return forward_call(*input, **kwargs)\n",
            "  File \"/content/SuperGluePretrainedNetwork/models/matching.py\", line 82, in forward\n",
            "    pred = {**pred, **self.superglue(data)}\n",
            "  File \"/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py\", line 1130, in _call_impl\n",
            "    return forward_call(*input, **kwargs)\n",
            "  File \"/content/SuperGluePretrainedNetwork/models/superglue.py\", line 253, in forward\n",
            "    desc0, desc1 = self.gnn(desc0, desc1)\n",
            "  File \"/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py\", line 1130, in _call_impl\n",
            "    return forward_call(*input, **kwargs)\n",
            "  File \"/content/SuperGluePretrainedNetwork/models/superglue.py\", line 138, in forward\n",
            "    delta0, delta1 = layer(desc0, src0), layer(desc1, src1)\n",
            "  File \"/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py\", line 1130, in _call_impl\n",
            "    return forward_call(*input, **kwargs)\n",
            "  File \"/content/SuperGluePretrainedNetwork/models/superglue.py\", line 120, in forward\n",
            "    message = self.attn(x, source, source)\n",
            "  File \"/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py\", line 1130, in _call_impl\n",
            "    return forward_call(*input, **kwargs)\n",
            "  File \"/content/SuperGluePretrainedNetwork/models/superglue.py\", line 108, in forward\n",
            "    x, _ = attention(query, key, value)\n",
            "  File \"/content/SuperGluePretrainedNetwork/models/superglue.py\", line 89, in attention\n",
            "    scores = torch.einsum('bdhn,bdhm->bhnm', query, key) / dim**.5\n",
            "  File \"/usr/local/lib/python3.7/dist-packages/torch/functional.py\", line 360, in einsum\n",
            "    return _VF.einsum(equation, operands)  # type: ignore[attr-defined]\n",
            "KeyboardInterrupt\n",
            "DNCNN mean value: nan\n"
          ]
        },
        {
          "ename": "FileNotFoundError",
          "evalue": "ignored",
          "output_type": "error",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mFileNotFoundError\u001b[0m                         Traceback (most recent call last)",
            "\u001b[0;32m<timed exec>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n",
            "\u001b[0;32m<ipython-input-9-7ed57ee706c6>\u001b[0m in \u001b[0;36mdownload_folder_in_zip\u001b[0;34m(dir_to_zip, output_filename, delete_dir_after_download)\u001b[0m\n\u001b[1;32m      3\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mdelete_dir_after_download\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m         \u001b[0mos\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msystem\u001b[0m\u001b[0;34m(\u001b[0m \u001b[0;34m\"rm -r {}\"\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mformat\u001b[0m\u001b[0;34m(\u001b[0m \u001b[0mdir_to_zip\u001b[0m \u001b[0;34m)\u001b[0m \u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 5\u001b[0;31m     \u001b[0mfiles\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdownload\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0moutput_filename\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      6\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      7\u001b[0m \u001b[0;32mdef\u001b[0m \u001b[0mmake_path_txt\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mTRUTH_DATASET\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mRESTORED_DATASET\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mnum_elements\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;36m100\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/google/colab/files.py\u001b[0m in \u001b[0;36mdownload\u001b[0;34m(filename)\u001b[0m\n\u001b[1;32m    179\u001b[0m   \u001b[0;32mif\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0m_os\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpath\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mexists\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfilename\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    180\u001b[0m     \u001b[0mmsg\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m'Cannot find file: {}'\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mformat\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfilename\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 181\u001b[0;31m     \u001b[0;32mraise\u001b[0m \u001b[0mFileNotFoundError\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmsg\u001b[0m\u001b[0;34m)\u001b[0m  \u001b[0;31m# pylint: disable=undefined-variable\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    182\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    183\u001b[0m   \u001b[0mcomm_manager\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_IPython\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mget_ipython\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mkernel\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcomm_manager\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mFileNotFoundError\u001b[0m: Cannot find file: superpoint_DNCNN.zip"
          ]
        }
      ],
      "source": [
        "%%time\n",
        "# DNCNN\n",
        "make_path_txt('ground_truth_color', 'results_DNCNN3')\n",
        "\n",
        "!python match_pairs.py --input_pairs /content/pairs.txt --output_dir /content/superpoint_DNCNN --nms_radius 1 --resize -1 --max_keypoints 1000 --viz\n",
        "\n",
        "match_values_d = []\n",
        "path = '/content/superpoint_DNCNN'\n",
        "for pair in os.listdir(path):\n",
        "    npz = np.load(path+'/'+pair)\n",
        "    match_values_d.append(np.sum(npz['matches']>-1))\n",
        "print('DNCNN mean value:', np.mean(np.array(match_values_d)))\n",
        "download_folder_in_zip('/content/results_match_DNCNN', 'superpoint_DNCNN.zip')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "S-aFpWj9F_ju",
        "outputId": "f5ad7fc3-414c-4d9c-b989-50c551531664"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Namespace(cache=False, eval=False, fast_viz=False, force_cpu=False, input_dir='assets/scannet_sample_images/', input_pairs='/content/pairs.txt', keypoint_threshold=0.005, match_threshold=0.1, max_keypoints=1000, max_length=-1, nms_radius=1, opencv_display=False, output_dir='/content/superpoint_FBCNN', resize=[-1], resize_float=False, show_keypoints=False, shuffle=False, sinkhorn_iterations=20, superglue='indoor', viz=True, viz_extension='png')\n",
            "Will not resize images\n",
            "Running inference on device \"cpu\"\n",
            "Loaded SuperPoint model\n",
            "Loaded SuperGlue model (\"indoor\" weights)\n",
            "Looking for data in directory \"assets/scannet_sample_images\"\n",
            "Will write matches to directory \"/content/superpoint_FBCNN\"\n",
            "Will write visualization images to directory \"/content/superpoint_FBCNN\"\n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair     0 of   100] load_image=0.265 matcher=32.087 viz_match=2.314 total=34.666 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair     1 of   100] load_image=0.251 matcher=30.501 viz_match=2.286 total=33.039 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair     2 of   100] load_image=0.238 matcher=29.962 viz_match=2.282 total=32.482 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair     3 of   100] load_image=0.233 matcher=29.890 viz_match=2.242 total=32.365 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair     4 of   100] load_image=0.238 matcher=29.380 viz_match=2.237 total=31.855 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair     5 of   100] load_image=0.233 matcher=29.090 viz_match=2.275 total=31.599 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair     6 of   100] load_image=0.228 matcher=28.440 viz_match=2.282 total=30.950 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair     7 of   100] load_image=0.226 matcher=28.072 viz_match=2.219 total=30.517 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair     8 of   100] load_image=0.238 matcher=27.894 viz_match=2.118 total=30.250 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair     9 of   100] load_image=0.234 matcher=27.630 viz_match=2.164 total=30.028 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    10 of   100] load_image=0.231 matcher=27.476 viz_match=2.114 total=29.822 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    11 of   100] load_image=0.226 matcher=28.120 viz_match=2.082 total=30.429 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    12 of   100] load_image=0.241 matcher=27.592 viz_match=2.032 total=29.865 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    13 of   100] load_image=0.236 matcher=27.254 viz_match=2.053 total=29.543 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    14 of   100] load_image=0.256 matcher=27.356 viz_match=2.074 total=29.686 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    15 of   100] load_image=0.252 matcher=26.938 viz_match=2.016 total=29.207 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    16 of   100] load_image=0.248 matcher=27.582 viz_match=2.197 total=30.027 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    17 of   100] load_image=0.237 matcher=27.572 viz_match=2.211 total=30.020 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    18 of   100] load_image=0.233 matcher=27.371 viz_match=2.167 total=29.772 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    19 of   100] load_image=0.242 matcher=27.386 viz_match=2.108 total=29.736 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    20 of   100] load_image=0.235 matcher=27.238 viz_match=2.158 total=29.631 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    21 of   100] load_image=0.232 matcher=27.173 viz_match=2.181 total=29.585 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    22 of   100] load_image=0.229 matcher=27.522 viz_match=2.130 total=29.880 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    23 of   100] load_image=0.240 matcher=27.041 viz_match=2.054 total=29.336 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    24 of   100] load_image=0.234 matcher=26.915 viz_match=2.102 total=29.251 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    25 of   100] load_image=0.230 matcher=27.312 viz_match=2.109 total=29.651 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    26 of   100] load_image=0.235 matcher=26.820 viz_match=2.045 total=29.100 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    27 of   100] load_image=0.229 matcher=26.733 viz_match=2.103 total=29.065 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    28 of   100] load_image=0.226 matcher=27.018 viz_match=2.167 total=29.410 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    29 of   100] load_image=0.227 matcher=27.308 viz_match=2.129 total=29.663 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    30 of   100] load_image=0.237 matcher=26.726 viz_match=2.039 total=29.003 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    31 of   100] load_image=0.230 matcher=26.682 viz_match=2.393 total=29.306 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    32 of   100] load_image=0.238 matcher=27.612 viz_match=2.344 total=30.194 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    33 of   100] load_image=0.230 matcher=27.395 viz_match=2.236 total=29.861 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    34 of   100] load_image=0.244 matcher=27.303 viz_match=2.209 total=29.756 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    35 of   100] load_image=0.238 matcher=27.122 viz_match=2.218 total=29.578 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    36 of   100] load_image=0.234 matcher=27.038 viz_match=2.199 total=29.472 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    37 of   100] load_image=0.236 matcher=26.741 viz_match=2.230 total=29.207 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    38 of   100] load_image=0.264 matcher=26.928 viz_match=2.241 total=29.433 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    39 of   100] load_image=0.250 matcher=26.962 viz_match=2.253 total=29.465 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    40 of   100] load_image=0.241 matcher=27.041 viz_match=2.332 total=29.614 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    41 of   100] load_image=0.280 matcher=26.871 viz_match=2.198 total=29.350 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    42 of   100] load_image=0.266 matcher=26.930 viz_match=2.217 total=29.413 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    43 of   100] load_image=0.253 matcher=26.973 viz_match=2.239 total=29.464 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    44 of   100] load_image=0.252 matcher=27.389 viz_match=2.169 total=29.810 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    45 of   100] load_image=0.260 matcher=26.975 viz_match=2.089 total=29.324 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    46 of   100] load_image=0.249 matcher=26.963 viz_match=2.133 total=29.344 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    47 of   100] load_image=0.241 matcher=27.452 viz_match=2.138 total=29.831 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    48 of   100] load_image=0.242 matcher=27.034 viz_match=2.064 total=29.340 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    49 of   100] load_image=0.236 matcher=26.963 viz_match=2.119 total=29.317 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    50 of   100] load_image=0.234 matcher=27.331 viz_match=2.155 total=29.719 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    51 of   100] load_image=0.230 matcher=27.263 viz_match=2.135 total=29.628 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    52 of   100] load_image=0.243 matcher=26.922 viz_match=2.059 total=29.224 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    53 of   100] load_image=0.243 matcher=27.379 viz_match=2.128 total=29.751 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    54 of   100] load_image=0.235 matcher=27.247 viz_match=2.175 total=29.657 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    55 of   100] load_image=0.234 matcher=27.166 viz_match=2.115 total=29.515 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    56 of   100] load_image=0.245 matcher=27.224 viz_match=2.045 total=29.514 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    57 of   100] load_image=0.237 matcher=27.009 viz_match=2.103 total=29.348 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    58 of   100] load_image=0.233 matcher=26.894 viz_match=2.117 total=29.244 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    59 of   100] load_image=0.238 matcher=26.916 viz_match=2.039 total=29.194 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    60 of   100] load_image=0.230 matcher=26.867 viz_match=2.105 total=29.202 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    61 of   100] load_image=0.232 matcher=26.883 viz_match=2.149 total=29.264 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    62 of   100] load_image=0.229 matcher=27.376 viz_match=2.121 total=29.725 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    63 of   100] load_image=0.242 matcher=26.927 viz_match=2.049 total=29.218 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    64 of   100] load_image=0.234 matcher=26.891 viz_match=2.110 total=29.234 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    65 of   100] load_image=0.230 matcher=27.312 viz_match=2.150 total=29.692 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    66 of   100] load_image=0.228 matcher=27.238 viz_match=2.095 total=29.561 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    67 of   100] load_image=0.239 matcher=26.872 viz_match=2.036 total=29.147 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    68 of   100] load_image=0.230 matcher=28.319 viz_match=2.085 total=30.634 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    69 of   100] load_image=0.230 matcher=28.406 viz_match=2.113 total=30.749 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    70 of   100] load_image=0.238 matcher=27.776 viz_match=2.048 total=30.062 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    71 of   100] load_image=0.233 matcher=28.042 viz_match=2.123 total=30.398 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    72 of   100] load_image=0.228 matcher=27.786 viz_match=2.177 total=30.191 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    73 of   100] load_image=0.225 matcher=27.648 viz_match=2.162 total=30.034 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    74 of   100] load_image=0.239 matcher=27.610 viz_match=2.075 total=29.925 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    75 of   100] load_image=0.231 matcher=27.434 viz_match=2.141 total=29.806 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    76 of   100] load_image=0.227 matcher=27.387 viz_match=2.172 total=29.786 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    77 of   100] load_image=0.231 matcher=27.487 viz_match=2.083 total=29.802 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    78 of   100] load_image=0.232 matcher=27.380 viz_match=2.157 total=29.768 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    79 of   100] load_image=0.228 matcher=27.316 viz_match=2.195 total=29.739 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    80 of   100] load_image=0.228 matcher=27.759 viz_match=2.136 total=30.122 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    81 of   100] load_image=0.238 matcher=27.289 viz_match=2.055 total=29.583 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    82 of   100] load_image=0.232 matcher=27.176 viz_match=2.140 total=29.548 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    83 of   100] load_image=0.229 matcher=27.557 viz_match=2.145 total=29.931 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    84 of   100] load_image=0.232 matcher=27.074 viz_match=2.066 total=29.372 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    85 of   100] load_image=0.227 matcher=27.019 viz_match=2.126 total=29.373 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    86 of   100] load_image=0.224 matcher=27.448 viz_match=2.165 total=29.837 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    87 of   100] load_image=0.221 matcher=27.311 viz_match=2.150 total=29.682 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    88 of   100] load_image=0.236 matcher=26.960 viz_match=2.068 total=29.264 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    89 of   100] load_image=0.233 matcher=28.772 viz_match=2.149 total=31.154 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    90 of   100] load_image=0.229 matcher=28.266 viz_match=2.194 total=30.689 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    91 of   100] load_image=0.226 matcher=27.925 viz_match=2.132 total=30.283 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    92 of   100] load_image=0.240 matcher=27.751 viz_match=2.059 total=30.050 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    93 of   100] load_image=0.235 matcher=27.480 viz_match=2.135 total=29.850 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    94 of   100] load_image=0.230 matcher=27.334 viz_match=2.142 total=29.706 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    95 of   100] load_image=0.236 matcher=27.303 viz_match=2.070 total=29.609 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    96 of   100] load_image=0.231 matcher=27.181 viz_match=2.146 total=29.559 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    97 of   100] load_image=0.229 matcher=27.103 viz_match=2.193 total=29.525 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    98 of   100] load_image=0.227 matcher=27.484 viz_match=2.161 total=29.872 sec {0.0 FPS} \n",
            "Warning: input resolution is very large, results may vary\n",
            "Warning: input resolution is very large, results may vary\n",
            "[Finished pair    99 of   100] load_image=0.240 matcher=27.870 viz_match=2.180 total=30.291 sec {0.0 FPS} \n"
          ]
        },
        {
          "ename": "ValueError",
          "evalue": "ignored",
          "output_type": "error",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mValueError\u001b[0m                                Traceback (most recent call last)",
            "\u001b[0;32m<timed exec>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/numpy/lib/npyio.py\u001b[0m in \u001b[0;36mload\u001b[0;34m(file, mmap_mode, allow_pickle, fix_imports, encoding)\u001b[0m\n\u001b[1;32m    443\u001b[0m             \u001b[0;31m# Try a pickle\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    444\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0mallow_pickle\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 445\u001b[0;31m                 raise ValueError(\"Cannot load file containing pickled data \"\n\u001b[0m\u001b[1;32m    446\u001b[0m                                  \"when allow_pickle=False\")\n\u001b[1;32m    447\u001b[0m             \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mValueError\u001b[0m: Cannot load file containing pickled data when allow_pickle=False"
          ]
        }
      ],
      "source": [
        "%%time\n",
        "# FBCNN\n",
        "make_path_txt('ground_truth_color', 'results_FBCNN')\n",
        "\n",
        "!python match_pairs.py --input_pairs /content/pairs.txt --output_dir /content/superpoint_FBCNN --nms_radius 1 --resize -1 --max_keypoints 1000 --viz --match_threshold 0.1\n",
        "\n",
        "match_values_f = []\n",
        "path = '/content/superpoint_FBCNN'\n",
        "for pair in os.listdir(path):\n",
        "    npz = np.load(path+'/'+pair)\n",
        "    match_values_f.append(np.sum(npz['matches']>-1))\n",
        "print('FBCNN mean value:', np.mean(np.array(match_values_f)))\n",
        "download_folder_in_zip('/content/superpoint_FBCNN', 'superpoint_FBCNN.zip')"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4WXGsDis1weY"
      },
      "source": [
        "###### visualization"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "X1Ik_hQO7lWQ",
        "outputId": "719a23f7-f7da-4788-f022-c34011990ab9"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "ARCNN mean: 5938.17\n",
            "DNCNN mean: 6090.71\n",
            "FBCNN mean: 6133.46\n"
          ]
        }
      ],
      "source": [
        "print('ARCNN mean:', np.mean(np.array(match_values_a)))\n",
        "print('DNCNN mean:', np.mean(np.array(match_values_d)))\n",
        "print('FBCNN mean:', np.mean(np.array(match_values_f)))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 369
        },
        "id": "fG9UQkQ_1v_J",
        "outputId": "29588b0f-b691-4bd1-9231-e61ee0d11d6a"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 431.125x360 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "data = pd.DataFrame({'ARCNN': match_values_a})\n",
        "displot = sns.displot(data, bins = 20)\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 369
        },
        "id": "dUkjkW_G7AdB",
        "outputId": "885655fc-1638-4799-cdfe-f8e9a055f952"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 433.125x360 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "data = pd.DataFrame({'DNCNN': match_values_d})\n",
        "displot = sns.displot(data, bins = 20)\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 369
        },
        "id": "PepqSKZ37Ixf",
        "outputId": "f50af603-4421-4fe6-d33c-23f725157712"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 430.375x360 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "data = pd.DataFrame({'FBCNN': match_values_f})\n",
        "displot = sns.displot(data, bins = 20)\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 269
        },
        "id": "nQST13kr7LRE",
        "outputId": "97ef522c-d3e3-4c67-b32e-54e9438614fa"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "data = pd.DataFrame({'ARCNN': match_values_a, 'DNCNN3': match_values_d, 'FBCNN': match_values_f})\n",
        "sns.boxplot(data = data)\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7mmBjh8y8A9s"
      },
      "source": [
        "### coord matching"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "TLyG_P2I9pc-"
      },
      "source": [
        "###### running"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ISex1mGMogAC"
      },
      "source": [
        "featuring"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 4,
      "metadata": {
        "id": "MPKrET0aBVm9"
      },
      "outputs": [],
      "source": [
        "# ARCNN featuring\n",
        "\n",
        "# make_path_txt(TRUTH_DATASET, ARCNN_DATASET)\n",
        "\n",
        "# !python match_pairs.py --input_pairs /content/pairs.txt --output_dir /content/results_match_ARCNN --nms_radius 1 --resize -1 --max_keypoints 10000\n",
        "\n",
        "kp_tr = {}\n",
        "kp_a = {}\n",
        "path = 'superpoint_ARCNN_10000/content/superpoint_ARCNN'\n",
        "for i, pair in enumerate(os.listdir(path)):\n",
        "    npz = np.load(path+'/'+pair)\n",
        "    kp_tr[i] = npz['keypoints0']\n",
        "    kp_a[i] = npz['keypoints1']"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 5,
      "metadata": {
        "id": "MqJT_-sWCwF2"
      },
      "outputs": [],
      "source": [
        "# DNCNN featuring\n",
        "\n",
        "# make_path_txt(TRUTH_DATASET, DNCNN_DATASET)\n",
        "\n",
        "# !python match_pairs.py --input_pairs /content/pairs.txt --output_dir /content/results_match_DNCNN --nms_radius 1 --resize -1 --max_keypoints 10000\n",
        "\n",
        "# kp_tr = {}\n",
        "kp_d = {}\n",
        "path = 'superpoint_DNCNN_10000/content/superpoint_DNCNN'\n",
        "for i, pair in enumerate(os.listdir(path)):\n",
        "    npz = np.load(path+'/'+pair)\n",
        "    # kp_tr[i] = npz['keypoints0']\n",
        "    kp_d[i] = npz['keypoints1']"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "metadata": {
        "id": "fI0RfwpfCysK"
      },
      "outputs": [],
      "source": [
        "# FBCNN featuring\n",
        "# make_path_txt(TRUTH_DATASET, FBCNN_DATASET)\n",
        "# \n",
        "# !python match_pairs.py --input_pairs /content/pairs.txt --output_dir /content/results_match_FBCNN --nms_radius 1 --resize -1 --max_keypoints 10000\n",
        "\n",
        "# kp_tr = {}\n",
        "kp_f = {}\n",
        "path = 'superpoint_FBCNN_10000/content/superpoint_FBCNN'\n",
        "for i, pair in enumerate(os.listdir(path)):\n",
        "    npz = np.load(path+'/'+pair)\n",
        "    kp_tr[i] = npz['keypoints0']\n",
        "    kp_f[i] = npz['keypoints1']"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "z_kHrwEcohvK"
      },
      "source": [
        "matching"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "fTfIqf9ijvz3"
      },
      "outputs": [],
      "source": [
        "# ARCNN\n",
        "arcnn_coord_match = superpoint_coord_batch_match(kp_tr, kp_a)\n",
        "np.save('superpoint_arcnn_coord_match.npy', arcnn_coord_match)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "jjswY4EYkJPm",
        "outputId": "f8d1a47d-744e-4867-cc7a-fd26f928ef67"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "##################################################\n",
            "pair 0\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1929\n",
            "number of matches 2 VS 1: 1929\n",
            "cross matches: 1929\n",
            "cross match value: 0.1929\n",
            "\n",
            "##################################################\n",
            "pair 1\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1837\n",
            "number of matches 2 VS 1: 1837\n",
            "cross matches: 1837\n",
            "cross match value: 0.1837\n",
            "\n",
            "##################################################\n",
            "pair 2\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1831\n",
            "number of matches 2 VS 1: 1831\n",
            "cross matches: 1831\n",
            "cross match value: 0.1831\n",
            "\n",
            "##################################################\n",
            "pair 3\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1847\n",
            "number of matches 2 VS 1: 1847\n",
            "cross matches: 1847\n",
            "cross match value: 0.1847\n",
            "\n",
            "##################################################\n",
            "pair 4\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1833\n",
            "number of matches 2 VS 1: 1833\n",
            "cross matches: 1833\n",
            "cross match value: 0.1833\n",
            "\n",
            "##################################################\n",
            "pair 5\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1862\n",
            "number of matches 2 VS 1: 1862\n",
            "cross matches: 1862\n",
            "cross match value: 0.1862\n",
            "\n",
            "##################################################\n",
            "pair 6\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1833\n",
            "number of matches 2 VS 1: 1833\n",
            "cross matches: 1833\n",
            "cross match value: 0.1833\n",
            "\n",
            "##################################################\n",
            "pair 7\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1818\n",
            "number of matches 2 VS 1: 1818\n",
            "cross matches: 1818\n",
            "cross match value: 0.1818\n",
            "\n",
            "##################################################\n",
            "pair 8\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1851\n",
            "number of matches 2 VS 1: 1851\n",
            "cross matches: 1851\n",
            "cross match value: 0.1851\n",
            "\n",
            "##################################################\n",
            "pair 9\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1864\n",
            "number of matches 2 VS 1: 1864\n",
            "cross matches: 1864\n",
            "cross match value: 0.1864\n",
            "\n",
            "##################################################\n",
            "pair 10\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1904\n",
            "number of matches 2 VS 1: 1904\n",
            "cross matches: 1904\n",
            "cross match value: 0.1904\n",
            "\n",
            "##################################################\n",
            "pair 11\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1907\n",
            "number of matches 2 VS 1: 1907\n",
            "cross matches: 1907\n",
            "cross match value: 0.1907\n",
            "\n",
            "##################################################\n",
            "pair 12\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1933\n",
            "number of matches 2 VS 1: 1933\n",
            "cross matches: 1933\n",
            "cross match value: 0.1933\n",
            "\n",
            "##################################################\n",
            "pair 13\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1925\n",
            "number of matches 2 VS 1: 1925\n",
            "cross matches: 1925\n",
            "cross match value: 0.1925\n",
            "\n",
            "##################################################\n",
            "pair 14\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1860\n",
            "number of matches 2 VS 1: 1860\n",
            "cross matches: 1860\n",
            "cross match value: 0.186\n",
            "\n",
            "##################################################\n",
            "pair 15\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1972\n",
            "number of matches 2 VS 1: 1972\n",
            "cross matches: 1972\n",
            "cross match value: 0.1972\n",
            "\n",
            "##################################################\n",
            "pair 16\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1925\n",
            "number of matches 2 VS 1: 1925\n",
            "cross matches: 1925\n",
            "cross match value: 0.1925\n",
            "\n",
            "##################################################\n",
            "pair 17\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1848\n",
            "number of matches 2 VS 1: 1848\n",
            "cross matches: 1848\n",
            "cross match value: 0.1848\n",
            "\n",
            "##################################################\n",
            "pair 18\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1791\n",
            "number of matches 2 VS 1: 1791\n",
            "cross matches: 1791\n",
            "cross match value: 0.1791\n",
            "\n",
            "##################################################\n",
            "pair 19\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1899\n",
            "number of matches 2 VS 1: 1899\n",
            "cross matches: 1899\n",
            "cross match value: 0.1899\n",
            "\n",
            "##################################################\n",
            "pair 20\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1859\n",
            "number of matches 2 VS 1: 1859\n",
            "cross matches: 1859\n",
            "cross match value: 0.1859\n",
            "\n",
            "##################################################\n",
            "pair 21\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1902\n",
            "number of matches 2 VS 1: 1902\n",
            "cross matches: 1902\n",
            "cross match value: 0.1902\n",
            "\n",
            "##################################################\n",
            "pair 22\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1901\n",
            "number of matches 2 VS 1: 1901\n",
            "cross matches: 1901\n",
            "cross match value: 0.1901\n",
            "\n",
            "##################################################\n",
            "pair 23\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1842\n",
            "number of matches 2 VS 1: 1842\n",
            "cross matches: 1842\n",
            "cross match value: 0.1842\n",
            "\n",
            "##################################################\n",
            "pair 24\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1940\n",
            "number of matches 2 VS 1: 1940\n",
            "cross matches: 1940\n",
            "cross match value: 0.194\n",
            "\n",
            "##################################################\n",
            "pair 25\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1850\n",
            "number of matches 2 VS 1: 1850\n",
            "cross matches: 1850\n",
            "cross match value: 0.185\n",
            "\n",
            "##################################################\n",
            "pair 26\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1867\n",
            "number of matches 2 VS 1: 1867\n",
            "cross matches: 1867\n",
            "cross match value: 0.1867\n",
            "\n",
            "##################################################\n",
            "pair 27\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1870\n",
            "number of matches 2 VS 1: 1870\n",
            "cross matches: 1870\n",
            "cross match value: 0.187\n",
            "\n",
            "##################################################\n",
            "pair 28\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1846\n",
            "number of matches 2 VS 1: 1846\n",
            "cross matches: 1846\n",
            "cross match value: 0.1846\n",
            "\n",
            "##################################################\n",
            "pair 29\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1899\n",
            "number of matches 2 VS 1: 1899\n",
            "cross matches: 1899\n",
            "cross match value: 0.1899\n",
            "\n",
            "##################################################\n",
            "pair 30\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1919\n",
            "number of matches 2 VS 1: 1919\n",
            "cross matches: 1919\n",
            "cross match value: 0.1919\n",
            "\n",
            "##################################################\n",
            "pair 31\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1831\n",
            "number of matches 2 VS 1: 1831\n",
            "cross matches: 1831\n",
            "cross match value: 0.1831\n",
            "\n",
            "##################################################\n",
            "pair 32\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1869\n",
            "number of matches 2 VS 1: 1869\n",
            "cross matches: 1869\n",
            "cross match value: 0.1869\n",
            "\n",
            "##################################################\n",
            "pair 33\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1863\n",
            "number of matches 2 VS 1: 1863\n",
            "cross matches: 1863\n",
            "cross match value: 0.1863\n",
            "\n",
            "##################################################\n",
            "pair 34\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1894\n",
            "number of matches 2 VS 1: 1894\n",
            "cross matches: 1894\n",
            "cross match value: 0.1894\n",
            "\n",
            "##################################################\n",
            "pair 35\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1892\n",
            "number of matches 2 VS 1: 1892\n",
            "cross matches: 1892\n",
            "cross match value: 0.1892\n",
            "\n",
            "##################################################\n",
            "pair 36\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1823\n",
            "number of matches 2 VS 1: 1823\n",
            "cross matches: 1823\n",
            "cross match value: 0.1823\n",
            "\n",
            "##################################################\n",
            "pair 37\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1873\n",
            "number of matches 2 VS 1: 1873\n",
            "cross matches: 1873\n",
            "cross match value: 0.1873\n",
            "\n",
            "##################################################\n",
            "pair 38\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1836\n",
            "number of matches 2 VS 1: 1836\n",
            "cross matches: 1836\n",
            "cross match value: 0.1836\n",
            "\n",
            "##################################################\n",
            "pair 39\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1919\n",
            "number of matches 2 VS 1: 1919\n",
            "cross matches: 1919\n",
            "cross match value: 0.1919\n",
            "\n",
            "##################################################\n",
            "pair 40\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1938\n",
            "number of matches 2 VS 1: 1938\n",
            "cross matches: 1938\n",
            "cross match value: 0.1938\n",
            "\n",
            "##################################################\n",
            "pair 41\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1886\n",
            "number of matches 2 VS 1: 1886\n",
            "cross matches: 1886\n",
            "cross match value: 0.1886\n",
            "\n",
            "##################################################\n",
            "pair 42\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1871\n",
            "number of matches 2 VS 1: 1871\n",
            "cross matches: 1871\n",
            "cross match value: 0.1871\n",
            "\n",
            "##################################################\n",
            "pair 43\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1807\n",
            "number of matches 2 VS 1: 1807\n",
            "cross matches: 1807\n",
            "cross match value: 0.1807\n",
            "\n",
            "##################################################\n",
            "pair 44\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1860\n",
            "number of matches 2 VS 1: 1860\n",
            "cross matches: 1860\n",
            "cross match value: 0.186\n",
            "\n",
            "##################################################\n",
            "pair 45\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1863\n",
            "number of matches 2 VS 1: 1863\n",
            "cross matches: 1863\n",
            "cross match value: 0.1863\n",
            "\n",
            "##################################################\n",
            "pair 46\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n"
          ]
        }
      ],
      "source": [
        "# DNCNN нужно от 0 до 46\n",
        "dncnn_coord_match = superpoint_coord_batch_match(kp_tr, kp_d)\n",
        "np.save('superpoint_dncnn_coord_match.npy', dncnn_coord_match)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "lp9gKY0NkKq7",
        "outputId": "fc75058a-10d3-4b86-b5fa-e7be61ab0bbe"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "##################################################\n",
            "pair 0\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2068\n",
            "number of matches 2 VS 1: 2068\n",
            "cross matches: 2068\n",
            "cross match value: 0.2068\n",
            "\n",
            "##################################################\n",
            "pair 1\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2013\n",
            "number of matches 2 VS 1: 2013\n",
            "cross matches: 2013\n",
            "cross match value: 0.2013\n",
            "\n",
            "##################################################\n",
            "pair 2\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2113\n",
            "number of matches 2 VS 1: 2113\n",
            "cross matches: 2113\n",
            "cross match value: 0.2113\n",
            "\n",
            "##################################################\n",
            "pair 3\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2083\n",
            "number of matches 2 VS 1: 2083\n",
            "cross matches: 2083\n",
            "cross match value: 0.2083\n",
            "\n",
            "##################################################\n",
            "pair 4\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2082\n",
            "number of matches 2 VS 1: 2082\n",
            "cross matches: 2082\n",
            "cross match value: 0.2082\n",
            "\n",
            "##################################################\n",
            "pair 5\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2059\n",
            "number of matches 2 VS 1: 2059\n",
            "cross matches: 2059\n",
            "cross match value: 0.2059\n",
            "\n",
            "##################################################\n",
            "pair 6\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2024\n",
            "number of matches 2 VS 1: 2024\n",
            "cross matches: 2024\n",
            "cross match value: 0.2024\n",
            "\n",
            "##################################################\n",
            "pair 7\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2135\n",
            "number of matches 2 VS 1: 2135\n",
            "cross matches: 2135\n",
            "cross match value: 0.2135\n",
            "\n",
            "##################################################\n",
            "pair 8\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2014\n",
            "number of matches 2 VS 1: 2014\n",
            "cross matches: 2014\n",
            "cross match value: 0.2014\n",
            "\n",
            "##################################################\n",
            "pair 9\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2113\n",
            "number of matches 2 VS 1: 2113\n",
            "cross matches: 2113\n",
            "cross match value: 0.2113\n",
            "\n",
            "##################################################\n",
            "pair 10\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2072\n",
            "number of matches 2 VS 1: 2072\n",
            "cross matches: 2072\n",
            "cross match value: 0.2072\n",
            "\n",
            "##################################################\n",
            "pair 11\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2083\n",
            "number of matches 2 VS 1: 2083\n",
            "cross matches: 2083\n",
            "cross match value: 0.2083\n",
            "\n",
            "##################################################\n",
            "pair 12\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2013\n",
            "number of matches 2 VS 1: 2013\n",
            "cross matches: 2013\n",
            "cross match value: 0.2013\n",
            "\n",
            "##################################################\n",
            "pair 13\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2052\n",
            "number of matches 2 VS 1: 2052\n",
            "cross matches: 2052\n",
            "cross match value: 0.2052\n",
            "\n",
            "##################################################\n",
            "pair 14\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1976\n",
            "number of matches 2 VS 1: 1976\n",
            "cross matches: 1976\n",
            "cross match value: 0.1976\n",
            "\n",
            "##################################################\n",
            "pair 15\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1943\n",
            "number of matches 2 VS 1: 1943\n",
            "cross matches: 1943\n",
            "cross match value: 0.1943\n",
            "\n",
            "##################################################\n",
            "pair 16\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2053\n",
            "number of matches 2 VS 1: 2053\n",
            "cross matches: 2053\n",
            "cross match value: 0.2053\n",
            "\n",
            "##################################################\n",
            "pair 17\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2019\n",
            "number of matches 2 VS 1: 2019\n",
            "cross matches: 2019\n",
            "cross match value: 0.2019\n",
            "\n",
            "##################################################\n",
            "pair 18\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2078\n",
            "number of matches 2 VS 1: 2078\n",
            "cross matches: 2078\n",
            "cross match value: 0.2078\n",
            "\n",
            "##################################################\n",
            "pair 19\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2034\n",
            "number of matches 2 VS 1: 2034\n",
            "cross matches: 2034\n",
            "cross match value: 0.2034\n",
            "\n",
            "##################################################\n",
            "pair 20\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2012\n",
            "number of matches 2 VS 1: 2012\n",
            "cross matches: 2012\n",
            "cross match value: 0.2012\n",
            "\n",
            "##################################################\n",
            "pair 21\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2059\n",
            "number of matches 2 VS 1: 2059\n",
            "cross matches: 2059\n",
            "cross match value: 0.2059\n",
            "\n",
            "##################################################\n",
            "pair 22\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2069\n",
            "number of matches 2 VS 1: 2069\n",
            "cross matches: 2069\n",
            "cross match value: 0.2069\n",
            "\n",
            "##################################################\n",
            "pair 23\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2009\n",
            "number of matches 2 VS 1: 2009\n",
            "cross matches: 2009\n",
            "cross match value: 0.2009\n",
            "\n",
            "##################################################\n",
            "pair 24\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2037\n",
            "number of matches 2 VS 1: 2037\n",
            "cross matches: 2037\n",
            "cross match value: 0.2037\n",
            "\n",
            "##################################################\n",
            "pair 25\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2067\n",
            "number of matches 2 VS 1: 2067\n",
            "cross matches: 2067\n",
            "cross match value: 0.2067\n",
            "\n",
            "##################################################\n",
            "pair 26\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2012\n",
            "number of matches 2 VS 1: 2012\n",
            "cross matches: 2012\n",
            "cross match value: 0.2012\n",
            "\n",
            "##################################################\n",
            "pair 27\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2061\n",
            "number of matches 2 VS 1: 2061\n",
            "cross matches: 2061\n",
            "cross match value: 0.2061\n",
            "\n",
            "##################################################\n",
            "pair 28\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2129\n",
            "number of matches 2 VS 1: 2129\n",
            "cross matches: 2129\n",
            "cross match value: 0.2129\n",
            "\n",
            "##################################################\n",
            "pair 29\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1966\n",
            "number of matches 2 VS 1: 1966\n",
            "cross matches: 1966\n",
            "cross match value: 0.1966\n",
            "\n",
            "##################################################\n",
            "pair 30\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2008\n",
            "number of matches 2 VS 1: 2008\n",
            "cross matches: 2008\n",
            "cross match value: 0.2008\n",
            "\n",
            "##################################################\n",
            "pair 31\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2126\n",
            "number of matches 2 VS 1: 2126\n",
            "cross matches: 2126\n",
            "cross match value: 0.2126\n",
            "\n",
            "##################################################\n",
            "pair 32\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2019\n",
            "number of matches 2 VS 1: 2019\n",
            "cross matches: 2019\n",
            "cross match value: 0.2019\n",
            "\n",
            "##################################################\n",
            "pair 33\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2053\n",
            "number of matches 2 VS 1: 2053\n",
            "cross matches: 2053\n",
            "cross match value: 0.2053\n",
            "\n",
            "##################################################\n",
            "pair 34\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2012\n",
            "number of matches 2 VS 1: 2012\n",
            "cross matches: 2012\n",
            "cross match value: 0.2012\n",
            "\n",
            "##################################################\n",
            "pair 35\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2027\n",
            "number of matches 2 VS 1: 2027\n",
            "cross matches: 2027\n",
            "cross match value: 0.2027\n",
            "\n",
            "##################################################\n",
            "pair 36\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1982\n",
            "number of matches 2 VS 1: 1982\n",
            "cross matches: 1982\n",
            "cross match value: 0.1982\n",
            "\n",
            "##################################################\n",
            "pair 37\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2019\n",
            "number of matches 2 VS 1: 2019\n",
            "cross matches: 2019\n",
            "cross match value: 0.2019\n",
            "\n",
            "##################################################\n",
            "pair 38\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2099\n",
            "number of matches 2 VS 1: 2099\n",
            "cross matches: 2099\n",
            "cross match value: 0.2099\n",
            "\n",
            "##################################################\n",
            "pair 39\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2010\n",
            "number of matches 2 VS 1: 2010\n",
            "cross matches: 2010\n",
            "cross match value: 0.201\n",
            "\n",
            "##################################################\n",
            "pair 40\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2149\n",
            "number of matches 2 VS 1: 2149\n",
            "cross matches: 2149\n",
            "cross match value: 0.2149\n",
            "\n",
            "##################################################\n",
            "pair 41\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2099\n",
            "number of matches 2 VS 1: 2099\n",
            "cross matches: 2099\n",
            "cross match value: 0.2099\n",
            "\n",
            "##################################################\n",
            "pair 42\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2042\n",
            "number of matches 2 VS 1: 2042\n",
            "cross matches: 2042\n",
            "cross match value: 0.2042\n",
            "\n",
            "##################################################\n",
            "pair 43\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2014\n",
            "number of matches 2 VS 1: 2014\n",
            "cross matches: 2014\n",
            "cross match value: 0.2014\n",
            "\n",
            "##################################################\n",
            "pair 44\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2043\n",
            "number of matches 2 VS 1: 2043\n",
            "cross matches: 2043\n",
            "cross match value: 0.2043\n",
            "\n",
            "##################################################\n",
            "pair 45\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2051\n",
            "number of matches 2 VS 1: 2051\n",
            "cross matches: 2051\n",
            "cross match value: 0.2051\n",
            "\n",
            "##################################################\n",
            "pair 46\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1999\n",
            "number of matches 2 VS 1: 1999\n",
            "cross matches: 1999\n",
            "cross match value: 0.1999\n",
            "\n",
            "##################################################\n",
            "pair 47\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2041\n",
            "number of matches 2 VS 1: 2041\n",
            "cross matches: 2041\n",
            "cross match value: 0.2041\n",
            "\n",
            "##################################################\n",
            "pair 48\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2081\n",
            "number of matches 2 VS 1: 2081\n",
            "cross matches: 2081\n",
            "cross match value: 0.2081\n",
            "\n",
            "##################################################\n",
            "pair 49\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2054\n",
            "number of matches 2 VS 1: 2054\n",
            "cross matches: 2054\n",
            "cross match value: 0.2054\n",
            "\n",
            "##################################################\n",
            "pair 50\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1989\n",
            "number of matches 2 VS 1: 1989\n",
            "cross matches: 1989\n",
            "cross match value: 0.1989\n",
            "\n",
            "##################################################\n",
            "pair 51\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2050\n",
            "number of matches 2 VS 1: 2050\n",
            "cross matches: 2050\n",
            "cross match value: 0.205\n",
            "\n",
            "##################################################\n",
            "pair 52\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2121\n",
            "number of matches 2 VS 1: 2121\n",
            "cross matches: 2121\n",
            "cross match value: 0.2121\n",
            "\n",
            "##################################################\n",
            "pair 53\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2080\n",
            "number of matches 2 VS 1: 2080\n",
            "cross matches: 2080\n",
            "cross match value: 0.208\n",
            "\n",
            "##################################################\n",
            "pair 54\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2033\n",
            "number of matches 2 VS 1: 2033\n",
            "cross matches: 2033\n",
            "cross match value: 0.2033\n",
            "\n",
            "##################################################\n",
            "pair 55\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2102\n",
            "number of matches 2 VS 1: 2102\n",
            "cross matches: 2102\n",
            "cross match value: 0.2102\n",
            "\n",
            "##################################################\n",
            "pair 56\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1990\n",
            "number of matches 2 VS 1: 1990\n",
            "cross matches: 1990\n",
            "cross match value: 0.199\n",
            "\n",
            "##################################################\n",
            "pair 57\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2029\n",
            "number of matches 2 VS 1: 2029\n",
            "cross matches: 2029\n",
            "cross match value: 0.2029\n",
            "\n",
            "##################################################\n",
            "pair 58\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2033\n",
            "number of matches 2 VS 1: 2033\n",
            "cross matches: 2033\n",
            "cross match value: 0.2033\n",
            "\n",
            "##################################################\n",
            "pair 59\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2131\n",
            "number of matches 2 VS 1: 2131\n",
            "cross matches: 2131\n",
            "cross match value: 0.2131\n",
            "\n",
            "##################################################\n",
            "pair 60\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2011\n",
            "number of matches 2 VS 1: 2011\n",
            "cross matches: 2011\n",
            "cross match value: 0.2011\n",
            "\n",
            "##################################################\n",
            "pair 61\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2031\n",
            "number of matches 2 VS 1: 2031\n",
            "cross matches: 2031\n",
            "cross match value: 0.2031\n",
            "\n",
            "##################################################\n",
            "pair 62\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2027\n",
            "number of matches 2 VS 1: 2027\n",
            "cross matches: 2027\n",
            "cross match value: 0.2027\n",
            "\n",
            "##################################################\n",
            "pair 63\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2104\n",
            "number of matches 2 VS 1: 2104\n",
            "cross matches: 2104\n",
            "cross match value: 0.2104\n",
            "\n",
            "##################################################\n",
            "pair 64\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2108\n",
            "number of matches 2 VS 1: 2108\n",
            "cross matches: 2108\n",
            "cross match value: 0.2108\n",
            "\n",
            "##################################################\n",
            "pair 65\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2028\n",
            "number of matches 2 VS 1: 2028\n",
            "cross matches: 2028\n",
            "cross match value: 0.2028\n",
            "\n",
            "##################################################\n",
            "pair 66\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2098\n",
            "number of matches 2 VS 1: 2098\n",
            "cross matches: 2098\n",
            "cross match value: 0.2098\n",
            "\n",
            "##################################################\n",
            "pair 67\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1985\n",
            "number of matches 2 VS 1: 1985\n",
            "cross matches: 1985\n",
            "cross match value: 0.1985\n",
            "\n",
            "##################################################\n",
            "pair 68\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2047\n",
            "number of matches 2 VS 1: 2047\n",
            "cross matches: 2047\n",
            "cross match value: 0.2047\n",
            "\n",
            "##################################################\n",
            "pair 69\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1933\n",
            "number of matches 2 VS 1: 1933\n",
            "cross matches: 1933\n",
            "cross match value: 0.1933\n",
            "\n",
            "##################################################\n",
            "pair 70\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2034\n",
            "number of matches 2 VS 1: 2034\n",
            "cross matches: 2034\n",
            "cross match value: 0.2034\n",
            "\n",
            "##################################################\n",
            "pair 71\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2054\n",
            "number of matches 2 VS 1: 2054\n",
            "cross matches: 2054\n",
            "cross match value: 0.2054\n",
            "\n",
            "##################################################\n",
            "pair 72\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1996\n",
            "number of matches 2 VS 1: 1996\n",
            "cross matches: 1996\n",
            "cross match value: 0.1996\n",
            "\n",
            "##################################################\n",
            "pair 73\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2028\n",
            "number of matches 2 VS 1: 2028\n",
            "cross matches: 2028\n",
            "cross match value: 0.2028\n",
            "\n",
            "##################################################\n",
            "pair 74\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2068\n",
            "number of matches 2 VS 1: 2068\n",
            "cross matches: 2068\n",
            "cross match value: 0.2068\n",
            "\n",
            "##################################################\n",
            "pair 75\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1977\n",
            "number of matches 2 VS 1: 1977\n",
            "cross matches: 1977\n",
            "cross match value: 0.1977\n",
            "\n",
            "##################################################\n",
            "pair 76\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2005\n",
            "number of matches 2 VS 1: 2005\n",
            "cross matches: 2005\n",
            "cross match value: 0.2005\n",
            "\n",
            "##################################################\n",
            "pair 77\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1980\n",
            "number of matches 2 VS 1: 1980\n",
            "cross matches: 1980\n",
            "cross match value: 0.198\n",
            "\n",
            "##################################################\n",
            "pair 78\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2052\n",
            "number of matches 2 VS 1: 2052\n",
            "cross matches: 2052\n",
            "cross match value: 0.2052\n",
            "\n",
            "##################################################\n",
            "pair 79\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2019\n",
            "number of matches 2 VS 1: 2019\n",
            "cross matches: 2019\n",
            "cross match value: 0.2019\n",
            "\n",
            "##################################################\n",
            "pair 80\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2089\n",
            "number of matches 2 VS 1: 2089\n",
            "cross matches: 2089\n",
            "cross match value: 0.2089\n",
            "\n",
            "##################################################\n",
            "pair 81\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2089\n",
            "number of matches 2 VS 1: 2089\n",
            "cross matches: 2089\n",
            "cross match value: 0.2089\n",
            "\n",
            "##################################################\n",
            "pair 82\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1994\n",
            "number of matches 2 VS 1: 1994\n",
            "cross matches: 1994\n",
            "cross match value: 0.1994\n",
            "\n",
            "##################################################\n",
            "pair 83\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2046\n",
            "number of matches 2 VS 1: 2046\n",
            "cross matches: 2046\n",
            "cross match value: 0.2046\n",
            "\n",
            "##################################################\n",
            "pair 84\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2057\n",
            "number of matches 2 VS 1: 2057\n",
            "cross matches: 2057\n",
            "cross match value: 0.2057\n",
            "\n",
            "##################################################\n",
            "pair 85\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2120\n",
            "number of matches 2 VS 1: 2120\n",
            "cross matches: 2120\n",
            "cross match value: 0.212\n",
            "\n",
            "##################################################\n",
            "pair 86\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2030\n",
            "number of matches 2 VS 1: 2030\n",
            "cross matches: 2030\n",
            "cross match value: 0.203\n",
            "\n",
            "##################################################\n",
            "pair 87\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2018\n",
            "number of matches 2 VS 1: 2018\n",
            "cross matches: 2018\n",
            "cross match value: 0.2018\n",
            "\n",
            "##################################################\n",
            "pair 88\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2011\n",
            "number of matches 2 VS 1: 2011\n",
            "cross matches: 2011\n",
            "cross match value: 0.2011\n",
            "\n",
            "##################################################\n",
            "pair 89\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2086\n",
            "number of matches 2 VS 1: 2086\n",
            "cross matches: 2086\n",
            "cross match value: 0.2086\n",
            "\n",
            "##################################################\n",
            "pair 90\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2047\n",
            "number of matches 2 VS 1: 2047\n",
            "cross matches: 2047\n",
            "cross match value: 0.2047\n",
            "\n",
            "##################################################\n",
            "pair 91\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2062\n",
            "number of matches 2 VS 1: 2062\n",
            "cross matches: 2062\n",
            "cross match value: 0.2062\n",
            "\n",
            "##################################################\n",
            "pair 92\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1988\n",
            "number of matches 2 VS 1: 1988\n",
            "cross matches: 1988\n",
            "cross match value: 0.1988\n",
            "\n",
            "##################################################\n",
            "pair 93\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2071\n",
            "number of matches 2 VS 1: 2071\n",
            "cross matches: 2071\n",
            "cross match value: 0.2071\n",
            "\n",
            "##################################################\n",
            "pair 94\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2069\n",
            "number of matches 2 VS 1: 2069\n",
            "cross matches: 2069\n",
            "cross match value: 0.2069\n",
            "\n",
            "##################################################\n",
            "pair 95\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 1992\n",
            "number of matches 2 VS 1: 1992\n",
            "cross matches: 1992\n",
            "cross match value: 0.1992\n",
            "\n",
            "##################################################\n",
            "pair 96\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2029\n",
            "number of matches 2 VS 1: 2029\n",
            "cross matches: 2029\n",
            "cross match value: 0.2029\n",
            "\n",
            "##################################################\n",
            "pair 97\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2089\n",
            "number of matches 2 VS 1: 2089\n",
            "cross matches: 2089\n",
            "cross match value: 0.2089\n",
            "\n",
            "##################################################\n",
            "pair 98\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2071\n",
            "number of matches 2 VS 1: 2071\n",
            "cross matches: 2071\n",
            "cross match value: 0.2071\n",
            "\n",
            "##################################################\n",
            "pair 99\n",
            "\n",
            "radius value: 1\n",
            "number of features for img1: 10000\n",
            "number of features for img2: 10000\n",
            "number of matches 1 VS 2: 2104\n",
            "number of matches 2 VS 1: 2104\n",
            "cross matches: 2104\n",
            "cross match value: 0.2104\n",
            "\n"
          ]
        }
      ],
      "source": [
        "# FBCNN\n",
        "fbcnn_coord_match = superpoint_coord_batch_match(kp_tr, kp_f)\n",
        "np.save('superpoint_fbcnn_coord_match.npy', fbcnn_coord_match)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "crgV6SyIc3vb"
      },
      "source": [
        "###### visualisation"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "hlWrLQ7vmMrw",
        "outputId": "11c0920a-8889-40bd-84fb-cbe189a34654"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Drive already mounted at /content/gdrive; to attempt to forcibly remount, call drive.mount(\"/content/gdrive\", force_remount=True).\n"
          ]
        }
      ],
      "source": [
        "drive.mount('/content/gdrive')\n",
        "\n",
        "arcnn_coord_match = np.load('/content/gdrive/MyDrive/superpoint_arcnn_coord_match.npy')\n",
        "dncnn_coord_match = np.load('/content/gdrive/MyDrive/superpoint_coord_matches_dncnn.npy')\n",
        "fbcnn_coord_match = np.load('/content/gdrive/MyDrive/superpoint_matches_fbcnn.npy')"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "dncnn_coord_match = np.array([float(value) for value in dncnn_coord_match])\n",
        "fbcnn_coord_match = np.array([float(value) for value in fbcnn_coord_match])"
      ],
      "metadata": {
        "id": "MYo0PFuaDqNs"
      },
      "execution_count": 9,
      "outputs": []
    },
    {
      "cell_type": "code",
      "execution_count": 10,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "W3lR82GXcckw",
        "outputId": "9cd229ba-3e35-4d61-b8c4-f609f1e67c95"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "ARCNN mean value: 0.177323\n",
            "DNCNN mean value: 0.18750899999999998\n",
            "FBCNN mean value: 0.20457699999999998\n"
          ]
        }
      ],
      "source": [
        "print('ARCNN mean value:', np.mean(arcnn_coord_match))\n",
        "print('DNCNN mean value:', np.mean(dncnn_coord_match))\n",
        "print('FBCNN mean value:', np.mean(fbcnn_coord_match))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 12,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 369
        },
        "id": "z37Q-PGtet-o",
        "outputId": "9930a826-1e41-49b2-8224-ac0d1c735a62"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 461.625x360 with 1 Axes>"
            ],
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAcwAAAFgCAYAAAA2BUkTAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAYkUlEQVR4nO3dfbRddX3n8feHBLACDiqRIuSBDi5GVMzgFZ8VtGUCpaJTimS5FKbYVEdmxmFq1TqtM7q6Vm1HqEqXNBVKdSo+tVhUwDD1gapYDQwP4UmBgiQyEKAV0VYNfOePsyMnN+fc/JLcc85N7vu11l5379/+7b2/P8LNJ3ufffZOVSFJkma2x6QLkCRpV2BgSpLUwMCUJKmBgSlJUgMDU5KkBgsnXcBsWrFiRV1++eWTLkPS/JZJF6DR2K3OMO+///5JlyBJ2k3tVoEpSdKoGJiSJDUwMCVJamBgSpLUwMCUJKmBgSlJUgMDU5KkBgamJEkNDExJkhoYmJIkNTAwJUlqYGBKktTAwJQkqYGBKUlSAwNTc97iJUtJMpZp8ZKlkx6upDlqt3qBtHZP6+/+LmevuXUsxzrruMPHchxJux7PMCVJamBgSpLUwMCUJKmBgSlJUgMDU5KkBgamJEkNDExJkhoYmJIkNTAwJUlqYGBKktTAwJQkqYGBKUlSAwNTkqQGBqYkSQ0MTO2wcb2nUpLmAt+HqR02rvdU+o5KSXOBZ5iSJDUwMCVJamBgSpLUYGSfYSa5ADgRuK+qntm1fQLY/IHU/sA/VdXyAdveCfwAeATYVFVTo6pTkqQWo7zp50LgXOAjmxuq6jWb55O8D/j+DNsfW1X3j6w6SZK2w8gCs6quTLJs0Lr0vitwCvDyUR1fkqTZNKnPMF8C3FtV3xmyvoA1Sa5OsmqmHSVZlWRtkrUbN26c9UIlSYLJBeZK4KIZ1r+4qo4CjgfenOSlwzpW1eqqmqqqqUWLFs12nZIkARMIzCQLgX8PfGJYn6ra0P28D7gYOHo81UmSNNgkzjB/EbilqtYPWplknyT7bZ4HjgPWjbE+SZK2MrLATHIRcBVweJL1Sc7oVp3KtMuxSZ6a5NJu8UDgq0muA74JfL6qLh9VnZIktRjlXbIrh7SfPqDte8AJ3fwdwLNHVZckSTvCJ/1IktTAwJQkqYGBuRvyPZWSNPt8H+ZuyPdUStLs8wxTkqQGBqYkSQ0MTEmSGhiYkiQ1MDAlSWpgYEqS1MDAlCSpgYEpSVIDA1OSpAYGpiRJDQxMSZIaGJiSJDUwMCVJamBgSpLUwMCUJKmBgSlJUgMDU5KkBgamJEkNDExJkhoYmJIkNTAwJUlqYGBKktTAwJQkqYGBKUlSAwNTkqQGBqYkSQ0MTEmSGowsMJNckOS+JOv62v5Hkg1Jru2mE4ZsuyLJrUluS/L2UdUoSVKrUZ5hXgisGNB+TlUt76ZLp69MsgD4E+B44AhgZZIjRlinJEnbNLLArKorgQd3YNOjgduq6o6q+gnwceCkWS1OkqTtNInPMM9Mcn13yfaJA9YfDNzdt7y+axsoyaoka5Os3bhx42zXqvkme5Bk5NPiJUsnPVJJ22nhmI/3IeA9QHU/3wf8+s7ssKpWA6sBpqamamcL1DxXj3L2mltHfpizjjt85MeQNLvGeoZZVfdW1SNV9SjwZ/Quv063AVjct3xI1yZJ0sSMNTCTHNS3+Gpg3YBu3wKeluTQJHsBpwKXjKM+SZKGGdkl2SQXAccAByRZD7wLOCbJcnqXZO8EfrPr+1Tgw1V1QlVtSnIm8AVgAXBBVd04qjolSWoxssCsqpUDms8f0vd7wAl9y5cCW33lRJKkSfFJP5IkNTAwJUlqYGBKktTAwJQkqYGBKUlSAwNTkqQGBqYkSQ0MTEmSGhiYkiQ1MDAlSWpgYEqS1MDAlCSpgYEpSVIDA1OSpAYGpiRJDQxMSZIaGJiSJDUwMCVJamBgSpLUwMCUJKmBgSlJUgMDU5KkBgamJEkNDExJkhoYmJIkNTAwJUlqYGBKktTAwJQkqYGBKUlSAwNTkqQGBqYkSQ1GFphJLkhyX5J1fW1/lOSWJNcnuTjJ/kO2vTPJDUmuTbJ2VDVKktRqlGeYFwIrprVdATyzqo4Evg28Y4btj62q5VU1NaL6JElqNrLArKorgQenta2pqk3d4jeAQ0Z1fEmSZtMkP8P8deCyIesKWJPk6iSrZtpJklVJ1iZZu3HjxlkvUpIkmFBgJnknsAn4yyFdXlxVRwHHA29O8tJh+6qq1VU1VVVTixYtGkG1kiRNIDCTnA6cCLy2qmpQn6ra0P28D7gYOHpsBUqSNMBYAzPJCuC3gVdW1Y+G9NknyX6b54HjgHWD+kqSNC6j/FrJRcBVwOFJ1ic5AzgX2A+4ovvKyHld36cmubTb9EDgq0muA74JfL6qLh9VnZIktVg4qh1X1coBzecP6fs94IRu/g7g2aOqS5KkHeGTfiRJamBgSpLUwMCUJKmBgSlJUgMDU5KkBgamJEkNDExJkhoYmJIkNTAwJUlqYGBKktTAwJQkqYGBKUlSAwNTkqQGBqYkSQ0MTEmSGhiYkiQ1MDAlSWpgYEqS1MDAlCSpgYE5JouXLCXJWCZJ0uxbOOkC5ov1d3+Xs9fcOpZjnXXc4WM5jiTNJ55hSpLUwMCUJKlBU2AmeVFLmyRJu6vWM8wPNrZJkrRbmvGmnyQvAF4ILEpyVt+qJwALRlmYJElzybbukt0L2Lfrt19f+0PAyaMqSpKkuWbGwKyqrwBfSXJhVd01ppokSZpzWr+HuXeS1cCy/m2q6uWjKEqSpLmmNTA/BZwHfBh4ZHTlSJI0N7UG5qaq+tBIK5EkaQ5r/VrJZ5P8xyQHJXnS5mlbGyW5IMl9Sdb1tT0pyRVJvtP9fOKQbU/r+nwnyWmNdUqSNBKtgXka8Fbg68DV3bS2YbsLgRXT2t4O/G1VPQ342255C10Yvwt4HnA08K5hwSpJ0jg0XZKtqkN3ZOdVdWWSZdOaTwKO6eb/Avgy8LZpff4dcEVVPQiQ5Ap6wXvRjtQhSdLOagrMJK8f1F5VH9mBYx5YVfd08/8POHBAn4OBu/uW13dtkiRNROtNP8/tm38c8ArgGmBHAvNnqqqS1M7sI8kqYBXAkiVLdmZXkiQN1XpJ9j/1LyfZH/j4Dh7z3iQHVdU9SQ4C7hvQZwOPXbYFOITepdtBta0GVgNMTU3tVPhKkjTMjr7e64fADn2uCVxC7yYiup9/M6DPF4Djkjyxu9nnuK5NkqSJaP0M87PA5rO3BcDTgU82bHcRvTPFA5Ksp3fn6x8An0xyBnAXcErXdwp4Y1W9oaoeTPIe4Fvdrt69+QYgSZImofUzzP/VN78JuKuq1m9ro6paOWTVKwb0XQu8oW/5AuCCxvokSTspyauAi4GnV9UtXdsy4GbgVnov5FgLnFFVP02yJ/Ae4FeBHwA/pneCc1mSO4Grq+pXu/2cDJxYVacnOZ3e3+/Lq+r6bv26bv2dOzmGLwO/1WXKrGq6JNs9hP0Wem8seSLwk9kuRJLUkwUL1yepWZsWLNzmCU5nJfDV7me/26tqOfAseveUnNK1vwc4CHhmVR0FvIot32z1nCRHDDnWeuCdjXXNCa2XZE8B/ojejTcBPpjkrVX16RHWJknz06OPHLz0bZ/7n7O1u7vee+K7ttUnyb7Ai4Fjgc/S+whtC1X1SJJvAgcneTzwG8ChVfXjbv29bPlx3fvoheJrBxzyc8BLkxxeVbcOqWkFvbPZX+uWj6F39nhikg/R+wbHzwGfrqqt6k3ycFXt2833n+Euovd89M1frXhLVX1thv88QPtNP+8EnltVp1XV6+k9fed3G7eVJM19JwGXV9W3gQeSPGd6hySPo/cEtsuBw4DvVtVDM+zzk8BRSQ4bsO5R4A+B35lh+/8DPC/JPt3ya3jsGxrvrKop4EjgZUmOnGE/070fOKeqnkvvcvKHWzZqDcw9qqr/6x8PbMe2kqS5byWPhdHH2fKy7L9Oci1wL3DP5s8dGzxC7+rkO4as/xjw/CQDv3VRVZvohfOvJFkI/DKPfbPilCTXAP8XeAYw7NLvIL8InNuN6RLgCd0Z9oxab/q5PMkXeOzRdK8BLt2O4iRJc1T3/O6XA8/qHiazAKgkb+263F5Vy5McAHwtySvpnf0tSfKEbZxlfpReYK6bvqKqNiV5H1s/HrXfx4EzgQeBtVX1gy5gf4velc9/THIhvYfqbHWIvvn+9XsAz6+qf5nhuFuZ8SwxyWFJXlRVbwX+lN6p75HAVXQPC5Ak7fJOBj5aVUurallVLQb+AXhJf6equp/eCzPeUVU/As4H3p9kL4Aki5L82rRtfgqcA/zXIce+kN4Z36Ih678CHEXv89LNZ8BPoPc8gO8nORA4fsi29yZ5epI9gFf3ta8BfvZAniTLh2y/hW1dVv1j4CGAqvrrqjqrqs6id9vxH7ccQJI0562k9/d6v79i67tlAT4DPD7JS4D/DmwEbuq+FvI5usyY5nyGXNGsqp8AHwCeMmT9I91+j+9+UlXX0bsUewu9y7rDbth5e7fN14F7+tr/MzCV5PokNwFvHLL9FrZ1SfbAqrphwABuGPAWEknSbNhjwYaWO1u3Z38zra6qYwe0faBv8Zl97QU8u2/db3fT9O2X9c3/GHhq3/KF9M4s+4/Vf7zp+zqT3mXZ/rbTh/Q9pm/+08BW3+bozpRfM+x4w2wrMPefYd3Pbe/BJEnbVo9sOmTSNWhr27okuzbJb0xvTPIGei+RliRpXtjWGeZbgIuTvJbHAnKK3uORXj10K0mSdjMzBmb31IYXJjmWx65hf76qvjjyyiRJmkNa34f5JeBLI65FkqQ5y6f1SJLUwMCUJKmBgSlJUgMDU5KkBgamJEkNDExJkhoYmJIkNTAwJUlqYGBKktTAwJQkqYGBKUlSAwNTkqQGBqYkSQ0MTEmSGhiYkiQ1MDAlSWpgYEqS1MDAlCSpgYEpSVKDsQdmksOTXNs3PZTkLdP6HJPk+319fm/cdUqS1G/huA9YVbcCywGSLAA2ABcP6Pp3VXXiOGuTJGmYSV+SfQVwe1XdNeE6JEma0aQD81TgoiHrXpDkuiSXJXnGsB0kWZVkbZK1GzduHE2VkqR5b2KBmWQv4JXApwasvgZYWlXPBj4IfGbYfqpqdVVNVdXUokWLRlOsJGnem+QZ5vHANVV17/QVVfVQVT3czV8K7JnkgHEXKEnSZpMMzJUMuRyb5OeTpJs/ml6dD4yxNkmStjD2u2QBkuwD/BLwm31tbwSoqvOAk4E3JdkE/DNwalXVJGqVJAkmFJhV9UPgydPazuubPxc4d9x1SZI0zKTvkpUkaZdgYEqS1MDAlCSpgYEpSVIDA1OSpAYGpiRJDQxMSZIaGJiSJDUwMCVJamBgSpLUwMCUJKmBgSlJUgMDU5KkBgamJEkNDExJkhoYmJIkNTAwJUlqYGBKktTAwJQmIXuQZCzT4iVLJz1aabewcNIFSPNSPcrZa24dy6HOOu7wsRxH2t15hilJUgMDU5KkBgamJEkNDExJkhoYmJIkNTAwJUlqYGBKktTAwJQkqYGBKUlSAwNTkqQGBqYkSQ0mFphJ7kxyQ5Jrk6wdsD5JPpDktiTXJzlqEnVKkgSTf/j6sVV1/5B1xwNP66bnAR/qfkqSNHZz+ZLsScBHqucbwP5JDpp0UZKk+WmSgVnAmiRXJ1k1YP3BwN19y+u7ti0kWZVkbZK1Gzdu3O4iFi9ZOpZ3EkqSdm2TvCT74qrakOQpwBVJbqmqK7d3J1W1GlgNMDU1Vdu7/fq7vzuW9xL6TkJJ2rVN7AyzqjZ0P+8DLgaOntZlA7C4b/mQrk2SpLGbSGAm2SfJfpvngeOAddO6XQK8vrtb9vnA96vqnjGXKkkSMLlLsgcCF3ef7S0EPlZVlyd5I0BVnQdcCpwA3Ab8CPgPE6pVkqTJBGZV3QE8e0D7eX3zBbx5nHVJkjTMXP5aiSRJc4aBKUlSAwNTkqQGBqYkSQ0MTEmSGhiYkiQ1MDAlSWpgYEqS1MDAlCSpgYEpSVIDA1Pa3WWPsbzzdfGSpZMeqTRSk3wfpqRxqEd956s0CzzDlCSpgYEpSVIDA1OSpAYGpiRJDQxMSZIaGJiSJDUwMCVJamBgSpLUwMCUJKmBgSlJUgMDU5KkBgamJEkNDExJkhoYmJIkNTAwJc0O37up3Zzvw5Q0O3zvpnZznmFKktTAwJQkqYGBKUlSg7EHZpLFSb6U5KYkNyb5LwP6HJPk+0mu7abfG3edkiT1m8RNP5uA/1ZV1yTZD7g6yRVVddO0fn9XVSdOoD5JkrYy9jPMqrqnqq7p5n8A3AwcPO46JEnaHhP9DDPJMuDfAn8/YPULklyX5LIkzxhrYZIkTTOx72Em2Rf4K+AtVfXQtNXXAEur6uEkJwCfAZ42ZD+rgFUAS5YsGWHFkqT5bCJnmEn2pBeWf1lVfz19fVU9VFUPd/OXAnsmOWDQvqpqdVVNVdXUokWLRlq3JGn+msRdsgHOB26uqrOH9Pn5rh9JjqZX5wPjq1KSpC1N4pLsi4DXATckubZr+x1gCUBVnQecDLwpySbgn4FTq6omUKskScAEArOqvgpkG33OBc4dT0WSJG2bT/qRJKmBgSlJUgMDU5KkBgampF3LmF5U7cuqNZ0vkJa0axnTi6rBl1VrS55hSpLUwMCUJKmBgSlJUgMDU5KkBgamJEkNDExJkhoYmJIkNTAwJUlqYGBKktTAwJQkqYGBKUlSAwNTkqQGBqYkSQ0MTEmSGhiYkiQ1MDAlSWpgYEqS1MDAlCSpgYEpSVIDA1OSpAYGpiRJDQxMSZIaGJiSJDUwMCVJamBgSpLUwMCUJKmBgSlJUoOJBGaSFUluTXJbkrcPWL93kk906/8+ybLxVylJ0mPGHphJFgB/AhwPHAGsTHLEtG5nAP9YVYcB5wDvHW+VkiRtaRJnmEcDt1XVHVX1E+DjwEnT+pwE/EU3/2ngFUkyxholSdpCqmq8B0xOBlZU1Ru65dcBz6uqM/v6rOv6rO+Wb+/63D9gf6uAVd3i4cCtIx7CMAcAW9W3G5tv4wXHPB/Mxnjvr6oVs1GM5paFky5gZ1XVamD1pOtIsraqpiZdx7jMt/GCY54P5tt4tX0mcUl2A7C4b/mQrm1gnyQLgX8FPDCW6iRJGmASgfkt4GlJDk2yF3AqcMm0PpcAp3XzJwNfrHFfO5Ykqc/YL8lW1aYkZwJfABYAF1TVjUneDaytqkuA84GPJrkNeJBeqM51E78sPGbzbbzgmOeD+TZebYex3/QjSdKuyCf9SJLUwMCUJKmBgTlAw6P7XprkmiSbuu+Vbm4/Nsm1fdO/JHlVt+7Q7jF/t3WP/dtrnGPalhGN+cIk/9C3bvk4xzSTHR1vt+4Pk9yY5OYkH9j8UI0kz0lyQ7fPn7XPFSMa85e7fW7+M37KuMbTYifH/N4k67rpNX3tc/p3WSNUVU59E70bkW4HfgHYC7gOOGJan2XAkcBHgJOH7OdJ9G5Yeny3/Eng1G7+POBNkx7rGMZ84bC+u+p4gRcCX+v2sQC4CjimW/dN4PlAgMuA4yc91jGM+cvA1KTHN4Ix/zJwBb0bI/ehd3f/E7p1c/Z32Wm0k2eYW9vmo/uq6s6quh54dIb9nAxcVlU/6v41/nJ6j/mD3mP/XjX7pe+wWR/z6EqdFTsz3gIeR+8v4L2BPYF7kxxE7y/Ub1RV0fsLeHf5Mx445tGXvNN2ZsxHAFdW1aaq+iFwPbBiF/hd1ggZmFs7GLi7b3l917a9TgUu6uafDPxTVW3ayX2OyijGvNnvJ7k+yTlJ9t7RAmfZDo+3qq4CvgTc001fqKqbu+3X78g+x2QUY97sz7vLsb87xy5D78z/19fRC8jHJzkAOJbew1Tm+u+yRsjAHIHubONZ9L5rOi8MGfM7gH8DPJfe5dq3TaC0WZXkMODp9J5QdTDw8iQvmWxVo7WNMb+2qp4FvKSbXjeZKmdXVa0BLgW+Tu8fgVcBj0y0KE2cgbm1lkf3bcspwMVV9dNu+QFg/+4xfzu6z1EaxZipqnuq58fAn9O7RDYX7Mx4Xw18o6oerqqH6X1W+YJu+0N2cJ/jMIoxU1Ubup8/AD7G3Pkzhp38/7qqfr+qllfVL9H7XPrbzP3fZY2Qgbm1lkf3bctK+i5Ndp9pfYneZ3zQe+zf38xCrbNl1scMPzvrpLtM9ypg3SzUOht2ZrzfBV6WZGGSPYGXATdX1T3AQ0me34339ew+f8YDx9wtHwDQtZ/I3Pkzhp0Yc5IFSZ7czR9J78agNbvA77JGadJ3Hc3FCTiB3r8mbwfe2bW9G3hlN/9cep9d/JDevzhv7Nt2Gb1/ce4xbZ+/QO8uytuATwF7T3qcYxjzF4Eb6P0l+r+BfSc9zp0dL707L/8UuBm4CTi7b59T3VhvB86le5LWXJlme8z07h69mt4NMTcC7wcWTHqcszTmx3VjvQn4BrC8b59z+nfZaXSTj8aTJKmBl2QlSWpgYEqS1MDAlCSpgYEpSVIDA1OSpAYGpiRJDQxMSZIa/H8GLoj6SdWtxwAAAABJRU5ErkJggg==\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "data = pd.DataFrame({'ARCNN value': arcnn_coord_match})\n",
        "displot = sns.displot(data)\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 11,
      "metadata": {
        "id": "wnWpuWCRe0DV",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 369
        },
        "outputId": "fde6e351-f3c1-45a4-8751-30ff49c48e75"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 463.625x360 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "data = pd.DataFrame({'DNCNN value': dncnn_coord_match})\n",
        "displot = sns.displot(data)\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 13,
      "metadata": {
        "id": "fl2iRFDAe0f0",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 369
        },
        "outputId": "462b0cc1-d6ba-47f5-9f7c-62c87f285bd5"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 460.875x360 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "data = pd.DataFrame({'FBCNN value': fbcnn_coord_match})\n",
        "displot = sns.displot(data)\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 16,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        },
        "id": "jMWFXe5ciYJn",
        "outputId": "bca28d95-0cf9-45b8-e93a-2debfb5d9030"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "data = pd.DataFrame({'ARCNN': arcnn_coord_match, 'DNCNN3': dncnn_coord_match, 'FBCNN': fbcnn_coord_match})\n",
        "sns.boxplot(data = data)\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "CrR9X1PCsptI"
      },
      "source": [
        "### distance between matching descriptors"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_-BLhkBNRx3-",
        "outputId": "2b476b14-37d7-48fc-b4dd-7ae7d8adf26d"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Mounted at /content/gdrive\n"
          ]
        }
      ],
      "source": [
        "drive.mount('/content/gdrive')\n",
        "\n",
        "x_truth = np.load('/content/gdrive/MyDrive/superpoint_keypoints_X_truth.npy')\n",
        "y_truth = np.load('/content/gdrive/MyDrive/superpoint_keypoints_Y_truth.npy')\n",
        "des_truth = np.load('/content/gdrive/MyDrive/superpoint_descriptors_truth.npy')\n",
        "\n",
        "x_comp = np.load('/content/gdrive/MyDrive/superpoint_keypoints_X_comp.npy')\n",
        "y_comp = np.load('/content/gdrive/MyDrive/superpoint_keypoints_Y_comp.npy')\n",
        "des_comp = np.load('/content/gdrive/MyDrive/superpoint_descriptors_comp.npy')\n",
        "\n",
        "x_arcnn = np.load('/content/gdrive/MyDrive/superpoint_keypoints_X_arcnn.npy')\n",
        "y_arcnn = np.load('/content/gdrive/MyDrive/superpoint_keypoints_Y_arcnn.npy')\n",
        "des_arcnn = np.load('/content/gdrive/MyDrive/superpoint_descriptors_arcnn.npy')\n",
        "\n",
        "x_fbcnn = np.load('/content/gdrive/MyDrive/superpoint_keypoints_X_fbcnn.npy')\n",
        "y_fbcnn = np.load('/content/gdrive/MyDrive/superpoint_keypoints_Y_fbcnn.npy')\n",
        "des_fbcnn = np.load('/content/gdrive/MyDrive/superpoint_descriptors_fbcnn.npy')\n",
        "\n",
        "\n",
        "    #### TAKE ONLY ONE IMAGE ####\n",
        "\n",
        "# superpoint_kp_truth[0 or 1]  -->  from kp get coord_x or coord_y\n",
        "# superpoint_kp_truth[(i-1)*10000:i*10000]  -->  from kp get i-th descriptor\n",
        "\n",
        "superpoint_kp_truth = dict(enumerate(zip(x_truth[0], y_truth[0])))\n",
        "superpoint_kp_comp = dict(enumerate(zip(x_comp[0], y_comp[0])))\n",
        "superpoint_kp_arcnn = dict(enumerate(zip(x_arcnn[0], y_arcnn[0])))\n",
        "superpoint_kp_fbcnn = dict(enumerate(zip(x_fbcnn[0], y_fbcnn[0])))\n",
        "\n",
        "\n",
        "des_truth = dict(enumerate(des_truth[:10000]))\n",
        "des_comp = dict(enumerate(des_comp[:10000]))\n",
        "des_arcnn = dict(enumerate(des_arcnn[:10000]))\n",
        "des_fbcnn = dict(enumerate(des_fbcnn[:10000]))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "kHCq0Te2gEZW"
      },
      "outputs": [],
      "source": [
        "def get_des_dist_data(superpoint_kp_truth, superpoint_kp_arcnn, des1, des2, name, radius=1, to_norm=False):\n",
        "    matches = dist_between_matched_descriptors(superpoint_kp_truth, superpoint_kp_arcnn, radius=radius)\n",
        "\n",
        "    des_dist = []\n",
        "    for k, v in matches.items():\n",
        "        if k in des1:\n",
        "            v1 = des1[k]\n",
        "        else:\n",
        "            v1 = des2[k]\n",
        "        if v in des1:\n",
        "            v2 = des1[v]\n",
        "        else:\n",
        "            v2 = des2[v]\n",
        "\n",
        "        if to_norm:\n",
        "            v1 = normalize(v1.reshape(1, -1))[0]\n",
        "            v2 = normalize(v2.reshape(1, -1))[0]\n",
        "\n",
        "        s = 0\n",
        "        for coord in v1 - v2:\n",
        "            s += coord**2\n",
        "\n",
        "        s = (s/len(v1))**(0.5)\n",
        "        des_dist.append(s)\n",
        "\n",
        "    data = pd.DataFrame({f'{name} desc dist, rad={radius}, mean={round(np.mean(des_dist), 4)}': des_dist})\n",
        "    return data"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "k0c5z9zEi7_3",
        "outputId": "28f06d4c-60d7-463e-bd14-fbd35f8cbfed"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "ARCNN\n",
            "radius value: 1\n",
            "number of features: 10000\n",
            "number of matching features: 1391\n",
            "radius value: 2\n",
            "number of features: 10000\n",
            "number of matching features: 4176\n",
            "radius value: 10\n",
            "number of features: 10000\n",
            "number of matching features: 5804\n",
            "\n",
            " FBCNN\n",
            "radius value: 1\n",
            "number of features: 10000\n",
            "number of matching features: 1687\n",
            "radius value: 2\n",
            "number of features: 10000\n",
            "number of matching features: 4661\n",
            "radius value: 10\n",
            "number of features: 10000\n",
            "number of matching features: 6147\n",
            "\n",
            " Compressed\n",
            "radius value: 1\n",
            "number of features: 10000\n",
            "number of matching features: 636\n",
            "radius value: 2\n",
            "number of features: 10000\n",
            "number of matching features: 2040\n",
            "radius value: 10\n",
            "number of features: 10000\n",
            "number of matching features: 3425\n"
          ]
        }
      ],
      "source": [
        "print('ARCNN')\n",
        "arcnn_1 = get_des_dist_data(superpoint_kp_truth, superpoint_kp_arcnn, des_truth, des_arcnn, radius=1, name = 'arcnn')\n",
        "arcnn_2 = get_des_dist_data(superpoint_kp_truth, superpoint_kp_arcnn, des_truth, des_arcnn, radius=2, name = 'arcnn')\n",
        "arcnn_10 = get_des_dist_data(superpoint_kp_truth, superpoint_kp_arcnn, des_truth, des_arcnn, radius=10, name = 'arcnn')\n",
        "\n",
        "print('\\n', 'FBCNN')\n",
        "fbcnn_1 = get_des_dist_data(superpoint_kp_truth, superpoint_kp_fbcnn, des_truth, des_fbcnn, radius=1, name = 'fbcnn')\n",
        "fbcnn_2 = get_des_dist_data(superpoint_kp_truth, superpoint_kp_fbcnn, des_truth, des_fbcnn, radius=2, name = 'fbcnn')\n",
        "fbcnn_10 = get_des_dist_data(superpoint_kp_truth, superpoint_kp_fbcnn, des_truth, des_fbcnn, radius=10, name = 'fbcnn')\n",
        "\n",
        "print('\\n', 'Compressed')\n",
        "comp_1 = get_des_dist_data(superpoint_kp_truth, superpoint_kp_comp, des_truth, des_comp, radius=1, name = 'comp')\n",
        "comp_2 = get_des_dist_data(superpoint_kp_truth, superpoint_kp_comp, des_truth, des_comp, radius=2, name = 'comp')\n",
        "comp_10 = get_des_dist_data(superpoint_kp_truth, superpoint_kp_comp, des_truth, des_comp, radius=10, name = 'comp')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 502
        },
        "id": "isR7G50PkrJQ",
        "outputId": "4c1223d6-203a-454b-b250-ffe3d8e6cf11"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7feee26cd490>"
            ]
          },
          "execution_count": 12,
          "metadata": {},
          "output_type": "execute_result"
        },
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1080x576 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "descriptor_dist = pd.concat([arcnn_1, arcnn_10, fbcnn_1, fbcnn_10, comp_1, comp_10], axis=1)\n",
        "\n",
        "sns.set(rc={\"figure.figsize\":(15, 8)})\n",
        "sns.histplot(data=descriptor_dist)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 502
        },
        "id": "bahgMMNGGI1u",
        "outputId": "5310d574-1c9b-4905-eabd-477b38816fd7"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7feee1baf910>"
            ]
          },
          "execution_count": 13,
          "metadata": {},
          "output_type": "execute_result"
        },
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1080x576 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "descriptor_dist = pd.concat([arcnn_1, fbcnn_1, comp_1], axis=1)\n",
        "\n",
        "sns.set(rc={\"figure.figsize\":(15, 8)})\n",
        "sns.histplot(data=descriptor_dist)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 502
        },
        "id": "ZWYkVUiFZlzz",
        "outputId": "7edab412-079c-4aac-8858-b7dd730719fd"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7feee0e39410>"
            ]
          },
          "execution_count": 16,
          "metadata": {},
          "output_type": "execute_result"
        },
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1080x576 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "descriptor_dist = pd.concat([arcnn_2, fbcnn_2, comp_2], axis=1)\n",
        "\n",
        "sns.set(rc={\"figure.figsize\":(15, 8)})\n",
        "sns.histplot(data=descriptor_dist)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 502
        },
        "id": "EcUrKiWvTQbB",
        "outputId": "2c15d905-f8e3-43db-f294-8214993ee5fa"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7feee26c4e10>"
            ]
          },
          "execution_count": 14,
          "metadata": {},
          "output_type": "execute_result"
        },
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1080x576 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "descriptor_dist = pd.concat([arcnn_10, fbcnn_10, comp_10], axis=1)\n",
        "\n",
        "sns.set(rc={\"figure.figsize\":(15, 8)})\n",
        "sns.histplot(data=descriptor_dist)"
      ]
    }
  ],
  "metadata": {
    "colab": {
      "collapsed_sections": [
        "8ND-0DPLNIJ-",
        "CrR9X1PCsptI"
      ],
      "provenance": []
    },
    "gpuClass": "standard",
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}