This code has been tested with Python 3.6.3, numpy 1.18.3, python-opencv 4.2.0, scipy 1.4.1 .
It also need some useful libraries: filterpy, imageio(opotion).
- install darknet(the newest yolov4 version)
It has been test with CUDA10.1 and cudnn7.6. Detailed install steps are all in
readme.md
.
# 1. modify as your need CMakeLists.txt or Makefile
$ ./build.sh
# or
$ make
- test the tracker demo
$ python darknet_video.py
If you find our work useful in your research, please consider citing:
@project{information2020spring,
title={Tracker: Using Kalman Filtering to track objects},
author={Jin Yujie},
year={2020}
}