Home

Awesome

Zero-shot Single Image Restoration through Controlled Perturbation of Koschmieder's Model

This is the PyTorch implementation for our CVPR 2021 paper:

**Aupendu Kar, Sobhan Kanti Dhara, Debashis Sen, Prabir Kumar Biswas. Zero-shot Single Image Restoration through Controlled Perturbation of Koschmieder's Model. [Project Website] [PAPER]

Dependencies

Test Datasets and Results of Our Algorithm

All the processed datasets that are used in this paper and the output images of the proposed algorithm are given below:

  1. Image Dehazing (Table 1 of Main Paper) Google Drive
  2. Underwater Image Enhancement (Table 2 of Main Paper) Google Drive
  3. Lowlight Image Enhancement (Table 3 of Main Paper) Google Drive

Test Codes

  1. Run HazeZeroShot.py to perform single image dehazing
  2. Run UnderWaterZeroShot.py to perform underwater image enhancement
  3. Run LowLightZeroShot.py to perform lowlight image enhancement

Citation

@InProceedings{Kar_2021_CVPR,
  author = {Kar, Aupendu and Dhara, Sobhan Kanti and Sen, Debashis and Biswas, Prabir Kumar},
  title = {Zero-shot Single Image Restoration through Controlled Perturbation of Koschmieder’s Model},
  booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  month = {June},
  year = {2021},
  pages = {16205-16215}
}

Contact

Aupendu Kar: mailtoaupendu[at]gmail[dot]com