Visual-in-Visual: A Unified and Efficient Baseline for Image Restoration

⭐ If VIVNet is helpful for you, please help star this repo and cite our work. Thanks! 🤗

🆕 News

• 2026.05.23: 🔥 This repo is released

⚙️ Setup

See INSTALL.md for the installation of dependencies required to run VIVNet.

🔧 Training and Evaluation

Training instructions for different tasks are provided in the respective directories.

🔧 Results & Models

Visual results and pre-trained models are available:

Visual Results

Pre-trained Models

📓 Citation

Please cite us if our work is useful for your research.

@ARTICLE{vivnet,
  author={Cui, Yuning and Ren, Wenqi and Shi, Boxin and Knoll, Alois},
  journal={IEEE Transactions on Pattern Analysis and Machine Intelligence}, 
  title={Visual-in-Visual: A Unified and Efficient Baseline for Image Restoration}, 
  year={2026}
}

✉️ Contact

Should you have any question, please contact yuning.cui@in.tum.de