sthanhng/yoloface ? reverse-engineered prompt

Build me a Python app that detects faces with YOLOv3 and can run on a single image, a video file, or a webcam feed. I want it to draw boxes around each face, show confidence scores if that’s easy, and save the processed result into an output folder. Please make it work with OpenCV and the usual deep learning libraries, and include simple command line options like choosing the input source and output directory.

If needed, use the pretrained face model files from the repo and look up the current docs online for anything unclear. Keep it easy to run from the command line, and make sure the image, video, and webcam examples all work end to end.