Skip to content

Official PyTorch implementation of TransforMatcher: Match-to-Match Attention for Semantic Correspondence (CVPR 2022 Poster)

Notifications You must be signed in to change notification settings

wookiekim/transformatcher

Repository files navigation

TransforMatcher: Match-to-Match Attention for Semantic Correspondence

This is the official pytorch implementation of the paper "TransforMatcher: Match-to-Match Attention for Semantic Correspondence" by Seungwook Kim, Juhong Min and Minsu Cho. Implemented on Python 3.7 and PyTorch 1.7.0.

Check out our project [website] and the paper on [arXiv]!

Requirements

Conda environment settings:

conda create -n tfm python=3.7
conda activate tfm

conda install pytorch=1.7.0 torchvision cudatoolkit=10.2 -c pytorch
conda install -c anaconda requests
conda install -c anaconda scipy
conda install -c anaconda pandas
conda install -c conda-forge einops
conda install -c conda-forge albumentations
pip install tensorboardX
pip install rotary-embedding-torch
pip install -U albumentations

Training

python train.py --benchmark {spair, pfpascal}

Testing

Trained models will be made available soon.

python test.py --benchmark {spair, pfpascal, pfwillow} 
               --load 'path_to_trained_model'

BibTeX

If you find our code or paper to be useful for your research, please consider citing our work:

@inproceedings{swkim2022tfmatcher,
  title={TransforMatcher:Match-to-Match Attention for Semantic Correspondence},
  author={Kim, Seungwook and Min, Juhong and Cho, Minsu },
  booktitle = {Proceedings of the {IEEE/CVF} Conference on Computer Vision and Pattern Recognition (CVPR)},
  year      = {2022}
}

Contact

Seungwook Kim ([email protected])

Feel free to reach out to me!

About

Official PyTorch implementation of TransforMatcher: Match-to-Match Attention for Semantic Correspondence (CVPR 2022 Poster)

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages