Readme.md

Smoking calling demo

1、copy your images to => data/Images

2、copy xml file to => data/Annotations

images file name and annotations file name must same

3、run python voc_label.py

cd yolov5_calling_smoking
python voc_label.py

4 weights

yolov5 weights download here

link: https://pan.baidu.com/s/1QIi0XFE8zBVuuEczHi75ww 
code: 3131 

unzip and copy to weights (folder)

smoking_calling weights download here

link: https://pan.baidu.com/s/1nSL4V0_nmZCGR7sI7-MUQw 
code: 3131 

unzip and copy to runs (folder)

5、run train.py

python train.py --batch 16 --data data/smoke.yaml --cfg models/yolov5l.yaml --weights weights/yolov5l.pt --epochs 100

Note：
--batch is batch size, default is 16
--data  config your data
--cfg   config for yolov5
--weights weights of yolov5
--epochs 

6. run detect.py for demo

python detect.py --source 0 --weights runs/exp2052/weights/best.pt

Note:
--source 0 will load your 0 camera for video detect, if you pass a path of images, then will detect all the images 
-- weights load weigths of the model
-- output  when you pass an output, then all detect images will be save to the output