bluedream1121 / NeMF

Official code implementation of NeMF (NeurIPS'22)

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Neural Matching Fields: Implicit Representation of Matching Fields for Visual Correspondence (NeurIPS'22)

This is the implementation of the paper "Neural Matching Fields: Implicit Representation of Matching Fields for Visual Correspondence" by Sunghwan Hong, Jisu Nam, Seokju Cho, Susung Hong, Sangryul Jeon, Dongbo Min and Seungryong Kim.

For more information, check out the paper on [arXiv] and the [project page].
Training code will be updated soon...

Overall Architecture

Our model NeMF is illustrated below:

alt text

Environment Settings

git clone https://github.com/KU-CVLAB/NeMF.git 
cd NeMF

conda create -n NeMF python=3.8
conda activate NeMF

conda install pytorch==1.11.0 torchvision==0.12.0 torchaudio==0.11.0 cudatoolkit=11.3 -c pytorch
pip install -U scikit-image
pip install git+https://github.com/albumentations-team/albumentations
pip install tensorboardX termcolor timm tqdm requests pandas einops matplotlib

Inference

alt text

  • Download pre-trained weights on Link

Result on SPair-71k :

  CUDA_VISIBLE_DEVICES=0 python test.py --pretrained ./SPAIR-NEMF --pretrained_file_name model_best.pth --benchmark spair

Result on PF-Pascal :

  CUDA_VISIBLE_DEVICES=0 python test.py --pretrained ./PF-PASCAL-NEMF --pretrained_file_name model_best.pth --benchmark pfpascal

Result on PF-Willow :

  CUDA_VISIBLE_DEVICES=0 python test.py --pretrained ./PF-PASCAL-NEMF --pretrained_file_name model_best.pth --benchmark pfwillow

Visualization

alt text alt text

Acknowledgement

We borrow code from public projects (huge thanks to all the projects). We mainly borrow code from DHPF and CATs.

BibTeX

If you find this research useful, please consider citing:

@inproceedings{hong2022neural,
  title={Neural Matching Fields: Implicit Representation of Matching Fields for Visual Correspondence},
  author={Sunghwan Hong and Jisu Nam and Seokju Cho and Susung Hong and Sangryul Jeon and Dongbo Min and Seungryong Kim},
  year={2022}
}

About

Official code implementation of NeMF (NeurIPS'22)


Languages

Language:Python 100.0%