tf_YOLOv2 using pretrained module,can detect picture,webcam or video. Requirement python3 tensorflow 1.3,1.1 or 1.2 should be ok numpy opencv-python model file1file2 usage run yolov2.py,edit func:demo_video or picture for your need