[Paper]   [Project Page]  
Yunpeng Bai, Cairong Wang, Shuzhao Xie, Chao Dong, Chun Yuan, Zhi Wang
Tsinghua University; Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences; Shanghai AI Laboratory, Shanghai, China

🎥 Click here to watch the demo video

• Python >= 3.7 (Recommend to use Anaconda or Miniconda)
• PyTorch >= 1.7
• Option: NVIDIA GPU + CUDA
• Option: Linux

1. Clone repo

   git clone https://github.com/yunpeng1998/TextIR
   cd TextIR

2. Install dependent packages

   pip install -r requirements.txt

Inference

python ./inference/inference_textIR_sr.py

We provide the training codes for TextIR.

Procedures

1. Training dataset preparation: FFHQ

Modify the configuration file options/*.yml accordingly.

1. Training

python -m torch.distributed.launch --nproc_per_node=8 --master_port=22021 basicsr/train.py -opt options/train/*.yml --launcher pytorch

@article{bai2025textir,
    title={Textir: A simple framework for text-based editable image restoration},
    author={Bai, Yunpeng and Wang, Cairong and Xie, Shuzhao and Dong, Chao and Yuan, Chun and Wang, Zhi},
    journal={IEEE Transactions on Visualization and Computer Graphics},
    year={2025},
    publisher={IEEE}
}