Any-Object-Detection-API

Creating Flask api and object detection using yolo3. we have only these 3 things requirements:

1.Pretrained weights file (yolov3.weights)

https://pjreddie.com/media/files/yolov3.weights
copy it,in yolo_files/

2.classes names file (coco.names)

present in repo not require

3.yolov3_custum.cfg (yolov3_custom.cfg)

present in repo

requirements

pip install requirement.txt

Run the app.py

python app.py
localhost:5000/

Example Input

OUTPUT

how can update yolov3_custom.cfg(Not Require for this repo project already done)

1.subdivisons=16/32

if classes=80

2.max_batches=(classes*2000)

(min_batches=4000,max_batches = 500200)
Note: if classes=1 then max_batches=4000

3.steps=80% of max_batches,90% of max_batches

(ex. steps=3200,3600 if classes=1)
(steps=400000,450000 if classes =80)
( on 17 line)
4.filters=(classes+5)*3

filters= (80+5)*3=255

last give classes name in coco.names

then finish updating.ok