Example of MXNet YOLO3 based Deep-SORT implementation.
Running demo.py generates mis-detected images for finetuning YOLO3.
Deep-SORT implementation is mostly based on nwojke/deep_sort and Qidian213/deep_sort_yolov3
- Install dependencies
$ pip install -r requirements.txt
-
Copy your own video file into the folder to test
-
Run demo with BYO-video
$ python demo.py --src video.mp4 --out-dir images --fps 12- Output Images. Frame images will be stored into output directory along with mis-detected objects.
$ ls images