Detection and custom training process works better, is more accurate and has more planned features to do:
https://github.com/pythonlessons/TensorFlow-2.x-YOLOv3
More information:
https://pylessons.com/YOLOv3-custom-data/
and
https://pylessons.com/YOLOv3-custom-training/
How to use YOLO v3 with webcam:
https://pylessons.com/YOLOv3-WebCam/
In future will update readme file
Requirements:
TensorFlow 1.15
Keras 2.2.4
Numpy (all versions should work)
opencv-python (all versions should work)