PSRGAN

Official PyTorch implementation of the paper Infrared Image Super-Resolution via Transfer Learning and PSRGAN accepted in IEEE SPL.

🔔Yongsong HUANG's Homepage 📌

Introduction

Recent advances in single image super-resolution (SISR) demonstrate the power of deep learning for achieving better performance. Because it is costly to recollect the training data and retrain the model for infrared (IR) image super-resolution, the availability of only a few samples for restoring IR images presents an important challenge in the field of SISR. To solve this problem, we first propose the progressive super-resolution generative adversarial network (PSRGAN) that includes the main path and branch path. The depthwise residual block (DWRB) is used to represent the features of the IR image in the main path. Then, the novel shallow lightweight distillation residual block (SLDRB) is used to extract the features of the readily available visible image in the other path. Furthermore, inspired by transfer learning, we propose the multistage transfer learning strategy for bridging the gap between different high-dimensional feature spaces that can improve the PSRGAN performance. Finally, quantitative and qualitative evaluations of two public datasets show that PSRGAN can achieve better results compared to the SR methods.

Approach overview

Main results

Requirements and dependencies

• Python 3.7
• Pytorch 0.4.1
• CUDA Version 10.2
• TITAN X (Pascal)
• Win10

Dataset prepare

Please check my homepage.

Model

Pre-trained models can be downloaded from this site.

Evaluation

Creating a new folder named model_zoo is necessary, please check the log file for more information about the settings.

Setting up the following directory structure:

.
├── model_zoo                   
|   ├──75000_G         # X4
|   |——5000_G          # X2 

---

Run

  main_test_kdsrgan.py

Citation

@ARTICLE{9424970, 
author={Huang, Yongsong and Jiang, Zetao and Lan, Rushi and Zhang, 
Shaoqin and Pi, Kui}, 
journal={IEEE Signal Processing Letters}, 
title={Infrared Image Super-Resolution via Transfer Learning 
and PSRGAN}, 
year={2021}, 
volume={28}, 
number={}, 
pages={982-986}, 
doi={10.1109/LSP.2021.3077801}}

Contact

If you meet any problems, please describe them and contact me.

Impolite or anonymous emails are not welcome. There may be some difficulties for me to respond to the email without self-introduce. Thank you for understanding.

Acknowledgement

Thanks to Kai Zhang for his work.