GFRRN

KO | EN

gitlite — search

#javascript #python #hacktoberfest #rust #linux #docker #macos #nodejs #windows #awesome #cli #awesome-list #pytorch #security

(★ 9)

Official implementation for "GFRRN: Explore the Gaps in Single Image Reflection Removal"

GFRRN 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

<div align="center">
 
 ## [CVPR26] GFRRN: Explore the Gaps in Single Image Reflection Removal
 
 
 
 Yu Chen</a>1,2
 ·
 Zewei He</a>1,2
 ·
 Xingyu Liu</a>1,2
 ·
 Zixuan Chen</a>3
 ·
 Zheming Lu</a>1
 
 
 1Zhejiang University, 2Huanjiang Lab, 3The Chinese University of Hong Kong

 
<h4>
 <a href="https://arxiv.org/abs/2602.22695">arXiv</a>
 · 
 <a href="https://tabe-i.github.io/GFRRN-project">Project Page</a>
 · 
 <a href="https://huggingface.co/datasets/tabe-i/Gap-Free">Datasets</a>
 </h4>
</div>

![teaser_img](https://raw.githubusercontent.com/tabe-i/GFRRN/main/teaser.png)

### :memo: ToDo List
- [x] Release training codes.
- [x] Release inference codes and pre-trained model.
- [x] Release GF Dataset (we will continue to expand it).
- [ ] Release a lightweight version (i.e., GFRRN_s).

### :gear: Quick Start
**Installation**
```shell
# clone this repo
git clone https://github.com/tabe-i/GFRRN.git
cd GFRRN

# create environment
conda create -n GFRRN python=3.10
conda activate GFRRN
pip install -r requirements.txt
```
**Prepare Dataset and Pre-trained Model**
Download [GFRRN.pt](https://huggingface.co/tabe-i/GFRRN_weight/tree/main), [GFRRN_s.pt]() and [swin_large_o365_finetune.pth](https://huggingface.co/tabe-i/GFRRN_weight/tree/main). Put them in ```weights``` folder. Download datasets ([Google Drive](https://drive.google.com/file/d/1hFZItZAzAt-LnfNj-2phBRwqplDUasQy/view?usp=sharing) or [Baidu Disk](https://pan.baidu.com/s/15zlk5o_-kx3ruKj4KfOvtA?pwd=1231)) and put them in ```datasets``` folder.

**Evaluation**
```shell
# GFRRN
python eval_sirs.py --name gfrrn_eval --arch gfrrn --model gfrrn_model_sirs_lrm --dataset sirs_dataset  --size_rounded --test_nature --weight_path "./weights/GFRRN.pt" --backbone_weight_path "./weights/swin_large_o365_finetune.pth" --base_dir "./datasets"
# GFRRN_s
python eval_sirs.py --name gfrrn_small_eval --arch gfrrn_s --model gfrrn_model_sirs_lrm --dataset sirs_dataset  --size_rounded --test_nature --weight_path "./weights/GFRRN_s.pt" --backbone_weight_path "./weights/swin_large_o365_finetune.pth" --base_dir "./datasets"
```

**Training**
```shell
# GFRRN
python train_sirs.py --name gfrrn_train --arch gfrrn --model gfrrn_model_sirs_lrm --dataset sirs_dataset --loss losses --size_rounded --backbone_weight_path "./weights/swin_large_o365_finetune.pth" --base_dir "./datasets" --batchSize 1
# GFRRN_s
python train_sirs.py --name gfrrn_small_train --arch gfrrn_s --model gfrrn_model_sirs_lrm --dataset sirs_dataset --loss losses --size_rounded --backbone_weight_path "./weights/swin_large_o365_finetune.pth" --base_dir "./datasets" --batchSize 1
```

**Testing**
```shell
# GFRRN
python test_sirs.py --name gfrrn_test --arch gfrrn --model gfrrn_model_sirs_lrm --dataset sirs_dataset  --size_rounded --test_nature --weight_path "./weights/GFRRN.pt" --backbone_weight_path "./weights/swin_large_o365_finetune.pth" --test_dir "[YOUR_TESTING_DATA_DIR]"
# GFRRN_s
python test_sirs.py --name gfrrn_small_test --arch gfrrn_s --model gfrrn_model_sirs_lrm --dataset sirs_dataset  --size_rounded --test_nature --weight_path "./weights/GFRRN_s.pt" --backbone_weight_path "./weights/swin_large_o365_finetune.pth" --test_dir "[YOUR_TESTING_DATA_DIR]"
```

### :star2: Acknowledgements
This project is developped on the codebase of [DSIT](https://github.com/mingcv/DSIT). We appreciate their great work!

### :love_you_gesture: Citation
If you find our paper and repo are helpful for your research, please consider citing:
```BibTeX
@article{chen2026gfrrn,
      title={GFRRN: Explore the Gaps in Single Image Reflection Removal}, 
      author={Yu Chen and Zewei He and Xingyu Liu and Zixuan Chen and Zheming Lu},
      year={2026},
      eprint={2602.22695},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2602.22695}, 
}
```

### :handshake: Contact
If you have any questions, please contact yuchen2024@zju.edu.cn or zeweihe@zju.edu.cn

GFRRN

GFRRN 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

GFRRN

GFRRN 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README