MFT: Long-Term Tracking of Every Pixel

Project Page Official implementation of the MFT tracker from the paper:

Michal Neoral, Jonáš Šerých, Jiří Matas: "MFT: Long-Term Tracking of Every Pixel", WACV 2024

Please cite our paper, if you use any of this.

@inproceedings{neoral2024mft,
               title={{MFT}: Long-Term Tracking of Every Pixel},
               author={Neoral, Michal and {\v{S}}er{\'{y}}ch, Jon{\'{a}}{\v{s}} and Matas, Ji{\v{r}}{\'{i}}},
               journal={arXiv preprint arXiv:2305.12998},
               year={2023},
}

Install

poetry install

Run the demo

Simply running:

poetry run python demo.py

should produce a demo_out directory with two visualizations.

Training

See train.org

License

The demo video in demo_in was extracted from youtube.

This work is licensed under the Attribution-NonCommercial-ShareAlike 4.0 International license. The MFT/RAFT directory contains a modified version of RAFT, which is licensed under BSD-3-Clause license. Our modifications (OcclusionAndUncertaintyBlock and its integration in raft.py) are licensed again under the Attribution-NonCommercial-ShareAlike 4.0 International.

Acknowledgments

This work was supported by Toyota Motor Europe, by the Grant Agency of the Czech Technical University in Prague, grant No. SGS23/173/OHK3/3T/13, and by the Research Center for Informatics project CZ.02.1.01/0.0/0.0/16_019/0000765 funded by OP VVV.

yumion/MFT-ONNX