detecting pedestrians using YOLOv3 (supporting images and videos)
This repository is adapteded from https://github.com/eriklindernoren/PyTorch-YOLOv3.
Sometimes there will be FPs,to improve accuracy we can add a post CNN filter following the cropped proposal. More details are in post_cnn_train.py, post_cnn_detect.py and post_filter.py.
I write a video.py for video detection.( detect.py is for image folder)