InstructIR: High-Quality Image Restoration Following Human Instructions


Marcos V. Conde1,2, Gregor Geigle1, Radu Timofte1,
▶ 1Computer Vision Lab, University of Wuerzburg    ▶ 2Sony PlayStation, FTG

Paper arXiv Code Demo 🤗 Models 🤗

TL;DR

InstructIR takes as input an image and a human-written instruction for how to improve that image. The (single) neural model performs all-in-one image restoration. InstructIR achieves state-of-the-art results on several restoration tasks including image denoising, deraining, deblurring, dehazing, and (low-light) image enhancement. Test the model now on HF 🤗 InstructIR space
🚀 You can start with the demo tutorial from our github.

🔥🔥 More content comming soon: all test results, more model versions, tables, etc.

Examples of InstructIR



Demo App 🤗

User

I love this photo, could you remove the raindrops? please keep the content intact

InstructIR

Sure! let me remove the rain

User

Great! can you make it look stunning? like a professional photo

InstructIR

I will enhance it

User

amazing, thank you InstructIR

Citation BibTeX


    @misc{conde2024instructir,
      title={High-Quality Image Restoration Following Human Instructions}, 
      author={Marcos V. Conde, Gregor Geigle, Radu Timofte},
      year={2024},
      journal={arXiv preprint}
    }

Acknowledgement

This work was partly supported by the The Humboldt Foundation (AvH). Marcos Conde is also supported by Sony Interactive Entertainment, FTG. This work is inspired in InstructPix2Pix.
For any inquiries contact Marcos V. Conde: marcos.conde [at] uni-wuerzburg.de

This website is adapted from Nerfies and LLaVA, licensed under a Creative Commons Attribution-ShareAlike 4.0 International License. We thank the original authors of this website template.