Skip to content

Latest commit

 

History

History
32 lines (22 loc) · 1.04 KB

README.md

File metadata and controls

32 lines (22 loc) · 1.04 KB

libtorch-yolov3-Tracker

Integrate libtorch-yolov3 with tracking algorithm making this project more engineering. Just a framework.

Many features and details need to be added as required according to demand.Hope to give some help.

This project is inspired by the libtorch-yolov3. Thanks to the author.

Requirements

  1. LibTorch v1.0.0+
  2. Cuda
  3. OpenCV 3.2 + opencv_contrib

Opencv_contrib

The API below is used in this tracking algorithm, you should know, because there is no such API in the official code, I added it.

CV_WRAP bool updateKCF( const Mat& image, Rect2d& boundingBox ,double& maxVal);

Modified opencv_contrib[vitx] is here

Training

Train yolov3 in Darknet

you can download this weight[vitx] for test.

Run tracker

modify labels_name in main.cpp

./yolo-tracker   <your_own_cfg>  <your_own_weights>  <path_to_video>