Repository for the paper Underwater Image Super-Resolution using Deep Residual Multipliers (ICRA 2020). Pre-print.

Resources

• Proposed dataset: USR-248
• Proposed model: SRDRM and SRDRM-GAN for underwater image super-resolution
• Models in comparison: SRGAN, ESRGAN, EDSRGAN, ResNetSR, SRCNN, and DSRCNN
• Requirements: TensorFlow >= 1.11 and Keras >= 2.2

Usage

• Download the data, setup data-paths in the training scripts
  • train-GAN-nx.py: SRDRM-GAN, SRGAN, ESRGAN, EDSRGAN
  • train-generative-models-nx.py: SRDRM, ResNetSR, SRCNN, DSRCNN
• Use the test-scripts for evaluating different models
  • A few test images: data/test/ (ground-truth: high_res)
• Use the measure.py for quantitative analysis

Bibliography Entry

@inproceedings{islam2020srdrm,
  title={{Underwater Image Super-Resolution using Deep Residual Multipliers}},
  author={Islam, Md Jahidul and Enan, Sadman Sakib and Luo, Peigen and Sattar, Junaed},
  booktitle={IEEE International Conference on Robotics and Automation (ICRA)},
  year={2020},
  organization={IEEE}
}

Acknowledgements

• https://github.com/Mulns/SuperSR
• https://github.com/david-gpu/srez
• https://github.com/wandb/superres
• https://github.com/tensorlayer/srgan
• https://github.com/icpm/super-resolution
• https://github.com/alexjc/neural-enhance
• https://github.com/jiny2001/dcscn-super-resolution
• https://github.com/titu1994/Image-Super-Resolution
• https://github.com/PacktPublishing/Advanced-Deep-Learning-with-Keras