It is an object detecion algorithm based on the YOLOv3 and a fast RCNN which prevents adversarial attacks on normal face detection techniques. This is developed with motive of providing better and more accurate surveillance solutions. Scroll down below to see how you can implement this on your own computer.

conda env create -f conda-cpu.yml

conda activate yolov3-cpu

python load_weights.py (a .tf file should be created)

python detect_video.py --video 0