Self-learning dataset generation:

# class 0 for T or 1 for CT
python vid.py --vid ./videos/t_model_0.mp4 --cls 0 --ml True

Train on generated cs:go dataset with:

python yolov5/train.py --img 640 --batch 10 --epochs 500 --data dataset.yaml --weights yolov5/yolov5s.pt

Inference:

python yolov5/detect.py --img 640 --conf 0.2  --weights ./data/weights/best.pt --source ./data/images

Inference on live gameplay (WIP):

python detect.py