Lit-Net

Harnessing Multi-resolution and Multi-scale Attention for Underwater Image Restoration.

• Currently, the paper is submitted to The Visual Computer.

• This paper deals with the underwater image restoration.
• For this, we have considered two of the main low-level vision tasks,
  • Underwater image enhancement,
  • Underwater image super-resolution.

Dataset

• For underwater image enhancement
  • EUVP,
  • UIEB, and
  • SUIM-E.
• For super-resolution,
  • UFO-120.

Rrequirements as given below.

Python 3.5.2
Pytorch '1.0.1.post2'
torchvision 0.2.2
opencv 4.0.0
scipy 1.2.1
numpy 1.16.2
tqdm

Checkpoints

• UIEB
• LSUI

Training

• Use the below command for training:

python train.py --checkpoints_dir --batch_size --learning_rate             

Testing

• Use the below command for testing:

python test.py  

For Underwater Semantic Segmentation

• To generate segmentation maps on enhanced images, follow SUIM.

Send us feedback

• If you have any queries or feedback, please contact us @([email protected]).

Citation

@misc{pramanick2024harnessingmultiresolutionmultiscaleattention,
      title={Harnessing Multi-resolution and Multi-scale Attention for Underwater Image Restoration}, 
      author={Alik Pramanick and Arijit Sur and V. Vijaya Saradhi},
      year={2024},
      eprint={2408.09912},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2408.09912}, 
}