Planar Object Tracking via Weighted Optical Flow

Project Page

Official implementation of the WOFT tracker from the paper

Jonáš Šerých, Jiří Matas: ”Planar Object Tracking via Weighted Optical Flow”, WACV 2023

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

@inproceedings{serych2023planar,
               title={Planar Object Tracking via Weighted Optical Flow},
               author={{\v{S}}er{\'{y}}ch, Jon{\'{a}}{\v{s}} and Matas, Ji{\v{r}}{\'{i}}},
	       booktitle={Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision},
               pages={1593--1602},
               year={2023}
}

License

This work is licensed under the Attribution-NonCommercial-ShareAlike 4.0 International license. The pytracking/external/RAFT directory contains a slightly modified copy of RAFT, which is licensed under BSD-3-Clause license, except for the pytracking/external/RAFT/raft_core/weighted_raft.py file, which is again licensed under the Attribution-NonCommercial-ShareAlike 4.0 International. The demo directory includes samples from the POT-210 dataset.

Installation

python -m pip install virtualenv
python -m virtualenv envs/WOFT
source envs/WOFT/bin/activate
python -m pip install tqdm einops opencv-python kornia==0.5.11 ipdb pandas matplotlib scipy

For the liteflownet experiment, install the following

python -m pip install torch==1.8.1 torchvision==0.9.1 tqdm einops opencv-python kornia==0.5.11 ipdb pandas matplotlib scipy openmim==0.2.1
mim install mmcv-full==1.6.1

# we use a very slightly modified / patched version of mmflow (included as submodule):
git submodule update --init --recursive
cd pytracking/external/mmflow
python -m pip install -v -e .

mim download mmflow --config liteflownet2_ft_4x1_600k_sintel_kitti_320x768

Demo

for video demo, run:

python WOFT_demo.py demo/V25_2/
# or
python WOFT_demo.py demo/V24_7/

Feel free to try other configs, like for example:

python WOFT_demo.py demo/V25_2/ --config pytracking/configs/WOFT_downscale_3x.py

Acknowledgments

This work was supported by Toyota Motor Europe, by CTU student grant SGS20/171/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.