MatchFlow

Website | Paper

Rethinking Optical Flow from Geometric Matching Consistent Perspective
Qiaole Dong, Chenjie Cao, Yanwei Fu
CVPR 2023

Requirements

The code has been tested with PyTorch 1.10.1 and Cuda 11.3.

conda create --name matchflow python=3.6
conda activate matchflow
pip install torch==1.10.1+cu113 torchvision==0.11.2+cu113 torchaudio==0.10.1 -f https://download.pytorch.org/whl/cu113/torch_stable.html
pip install matplotlib imageio einops scipy opencv-python tensorboard yacs timm pytorch_lightning
cd QuadTreeAttention
python setup.py install
cd ../

Required Data

To evaluate/train MatchFlow, you will need to download the required datasets.

• FlyingChairs
• FlyingThings3D
• Sintel
• KITTI
• HD1K (optional)

By default datasets.py will search for the datasets in these locations. You can create symbolic links to wherever the datasets were downloaded in the datasets folder

├── datasets
    ├── Sintel
        ├── test
        ├── training
    ├── KITTI
        ├── testing
        ├── training
        ├── devkit
    ├── FlyingChairs_release
        ├── data
    ├── FlyingThings3D
        ├── frames_cleanpass
        ├── frames_finalpass
        ├── optical_flow

Evaluation

You can evaluate a trained model using main.py

bash evaluate.sh

Training

Stage 1: Our pre-trained FME is downloaded from QuadTreeAttention and can be found in ckpts.

Stage 2: We used the following training schedule in our paper (2 GPUs). Training logs will be written to the runs which can be visualized using tensorboard.

bash train_standard.sh

Reference

If you found our paper helpful, please consider citing:

@inproceedings{dong2023rethinking,
  title={Rethinking Optical Flow from Geometric Matching Consistent  Perspective},
  author={Dong, Qiaole and Cao, Chenjie and Fu, Yanwei},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  year={2023}
}

Acknowledgement

Thanks to previous open-sourced repo:

• RAFT
• GMA
• QuadTreeAttention