fall_down-detect-yolov4

Some details

• darknet-yolov4-install-tutorial
• fall_down-yolov4-weights download: BaiDuYunPan 提取码->(krqn)
• yolov4.conv.137 -> GoogleDriver download: yolov4.conv.137

How to use this code (test)?

0. installed darknet-yolov4, and put darknet_API.py into ./darknet
1. put cfg into ./darknet
2. download fall_down-yolov4's weight, and put it in backup folder
3. Call the darknet_API main function:

   from darknet_API import Detect
   detect = Detect(metaPath=r'./cfg/fall_down.data', configPath=r'./cfg/yolov4-fall_down.cfg',\
                   weightPath=r'./backup/yolov4-fall_down_best.weights',\
                   namesPath=r'./cfg/fall_down.names')
   image = cv2.imread(r'/home/Datasets/20200714085948.jpg', -1)
   draw_img = detect.predict_image(image, save_path='./pred.jpg')
   

How to train yolov4 in darknet (train)?

0. Convert VOC format data to YOLO format data
1. Configure file information such as cfg
2. Call the darknet command:

   ./darknet detector train cfg/fall_down.data cfg/yolov4-fall_down.cfg yolov4.conv.137 -gpus 0 -map -dont_show
   

demo

• ./result: fall_down-detect demos