Introduction

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

Quick Start

1. Install dependencies

$ pip install -r requirements.txt

2. Copy your own video file into the folder to test

3. Run demo with BYO-video

$ python demo.py --src video.mp4 --out-dir images --fps 12

4. Output Images. Frame images will be stored into output directory along with mis-detected objects.

$ ls images