Dog Breed Detection App

Introduction

The Dog Breed Detection App is a simple Streamlit application that uses a pre-trained machine learning model to predict the breed of a dog based on an uploaded image. Whether you're a dog enthusiast or just curious about your dog's breed, this app can provide you with an answer.

Features

• Breed Prediction: Upload a picture of your dog, and the app will predict the most likely breed from a list of 70 different breeds.
• Easy to Use: The user-friendly interface makes it simple for anyone to use.

Demo

You can see a live demo of the app here.

Installation

To run the Dog Breed Detection App locally, follow these steps:

1. Clone this repository:

   git clone https://github.com/harichselvamc/Dog_Breed_detection
   

2. Change the directory to the project folder:

    cd Dog_Breed_detection
   

3. Install the required dependencies using pip:

    pip install -r requirements.txt
   
   

4. Run the Streamlit app:

    streamlit run app.py
   

The app will open in your web browser, allowing you to upload dog images for classification.

Folder Structure

The model has been trained on a dataset organized in the following folder structure:

    ./dog
    │
    └── Afghan
    │   ├── 001.jpg
    │   ├── 002.jpg
    │   ...
    │
    └── African Wild Dog
    │   ├── 001.jpg
    │   ├── 002.jpg
    │   ...
    ...

Each breed has its own folder containing multiple images.

Model Details

The Dog Breed Detection App uses a pre-trained machine learning model based on the InceptionV3 architecture. The model has been fine-tuned on a dataset of 70 different dog breeds.