Awesome

Introduction

X-Super-Resolution is dedicated to presenting the research efforts of XPixel in the realm of image super-resolution. We are thrilled to share research papers and corresponding open-source code crafted by our team.

Super-resolution algorithms aim to reconstruct high-resolution images from low-resolution counterparts, preserving and enhancing important details.

Super-resolution has applications in various domains such as surveillance, medical imaging, satellite imagery, and digital entertainment. It enhances image and video quality, making it invaluable for tasks that require high levels of detail and accuracy.

Table of Contents

• Representative Work
  • SRCNN
  • RealESRGAN
• Blind SR
• Classic SR

Papers

<a name="representative"></a>Representative Work:fire::fire::fire:

• <a name="srcnn"></a>Learning a Deep Convolutional Network for Image Super-Resolution<br> Chao Dong, Chen Change Loy, Kaiming He, Xiaoou Tang<br> Accepted at ECCV'14<br> :scroll:paper :house:project<br>

  <details close> <summary>more</summary> We propose a deep learning method for single image super-resolution (SR). Our method directly learns an end-to-end mapping between the low/high-resolution images. The mapping is represented as a deep convolutional neural network (CNN) that takes the low-resolution image as the input and outputs the high-resolution one. <div align="center"> <img src="assets/srcnn.png" width="600"/> </div> </details>

• <a name="realesrgan"></a>Real-ESRGAN: Training Real-World Blind Super-Resolution with Pure Synthetic Data<br> Xintao Wang, Liangbin Xie, Chao Dong, Ying Shan<br> Accepted at ICCVW'21<br> :scroll:paper :computer:code<br>

  <details close> <summary>more</summary> In this work, we extend the powerful ESRGAN to a practical restoration application, which is trained with pure synthetic data. Specifically:
  1. A high-order degradation modeling process is introduced to better simulate complex real-world degradations.
  2. We also consider the common ringing and overshoot artifacts in the synthesis process.
  3. In addition, we employ a U-Net discriminator with spectral normalization to increase discriminator capability and stabilize the training dynamics.

  Extensive comparisons have shown its superior visual performance than prior works on various real datasets.

  <div align="center"> <img src="assets/realesrgan.png" width="600"/> </div> </details>

Blind SR

• Blind Image Super-Resolution: A Survey and Beyond<br> Anran Liu, Yihao Liu, Jinjin Gu, Yu Qiao, Chao Dong<br> Accepted at TPAMI'22<br> :scroll:paper

• Crafting Training Degradation Distribution for the Accuracy-Generalization Trade-off in Real-World Super-Resolution<br> Ruofan Zhang, Jinjin Gu, Haoyu Chen, Chao Dong, Yulun Zhang, Wenming Yang<br> Accepted at ICML'23<br> :scroll:paper

• DeSRA: Detect and Delete the Artifacts of GAN-based Real-World Super-Resolution Models<br> Liangbin Xie, Xintao Wang, Xiangyu Chen, Gen Li, Ying Shan, Jiantao Zhou, Chao Dong<br> Accepted at ICML'23<br> :scroll:paper :computer:code

• OSRT: Omnidirectional Image Super-Resolution with Distortion-aware Transformer<br> Fanghua Yu, Xintao Wang, Mingdeng Cao, Gen Li, Ying Shan, Chao Dong<br> Accepted at CVPR'23<br> :scroll:paper :computer:code

• Metric Learning based Interactive Modulation for Real-World Super-Resolution<br> Chong Mou, Yanze Wu, Xintao Wang, Chao Dong, Jian Zhang, Ying Shan<br> Accepted at ECCV'22<br> :scroll:paper :computer:code

• A Closer Look at Blind Super-Resolution: Degradation Models, Baselines, and Performance Upper Bounds<br> Wenlong Zhang, Guangyuan Shi, Yihao Liu, Chao Dong, Xiao-Ming Wu<br> Accepted at CVPRW'22<br> :scroll:paper :computer:code

• GCFSR: a Generative and Controllable Face Super Resolution Method Without Facial and GAN Priors<br> Jingwen He, Wu Shi, Kai Chen, Lean Fu, Chao Dong<br> Accepted at CVPR'22<br> :scroll:paper

• Reflash Dropout in Image Super-Resolution<br> Xiangtao Kong, Xina Liu, Jinjin Gu, Yu Qiao, Chao Dong<br> Accepted at CVPR'22<br> :scroll:paper

• Suppressing Model Overfitting for Image Super-Resolution Networks<br> Ruicheng Feng, Jinjin Gu, Yu Qiao, Chao Dong<br> Accepted at CVPRW'19<br> :scroll:paper

• Blind Super-Resolution With Iterative Kernel Correction<br> Jinjin Gu, Hannan Lu, Wangmeng Zuo, Chao Dong<br> Accepted at CVPR'19<br> :scroll:paper

• Unsupervised Image Super-Resolution Using Cycle-in-Cycle Generative Adversarial Networks<br> Yuan Yuan, Siyuan Liu, Jiawei Zhang, Yongbing Zhang, Chao Dong, Liang Lin<br> Accepted at CVPRW'18<br> :scroll:paper

Classic SR

• Activating More Pixels in Image Super-Resolution Transformer<br> Xiangyu Chen, Xintao Wang, Jiantao Zhou, Yu Qiao, Chao Dong<br> Accepted at CVPR'23<br> :scroll:paper :computer:code

• Efficient Image Super-Resolution using Vast-Receptive-Field Attention<br> Lin Zhou, Haoming Cai, Jinjin Gu, Zheyuan Li, Yingqi Liu, Xiangyu Chen, Yu Qiao, Chao Dong<br> Accepted at ECCVW'22<br> :scroll:paper :computer:code

• Blueprint Separable Residual Network for Efficient Image Super-Resolution<br> Zheyuan Li, Yingqi Liu, Xiangyu Chen, Haoming Cai, Jinjin Gu, Yu Qiao, Chao Dong<br> Accepted at CVPRW'22<br> :scroll:paper :computer:code

• RepSR: Training Efficient VGG-style Super-Resolution Networks with Structural Re-Parameterization and Batch Normalization<br> Xintao Wang, Chao Dong, Ying Shan<br> Accepted at ACM MM'22<br> :scroll:paper :computer:code

• ClassSR: A General Framework to Accelerate Super-Resolution Networks by Data Characteristic<br> Xiangtao Kong, Hengyuan Zhao, Yu Qiao, Chao Dong<br> Accepted at CVPR'21<br> :scroll:paper :computer:code

• RankSRGAN: Super Resolution Generative Adversarial Networks with Learning to Rank<br> Wenlong Zhang, Yihao Liu, Chao Dong, Yu Qiao<br> Accepted at TPAMI'21<br> :scroll:paper :computer:code

• Efficient Image Super-Resolution Using Pixel Attention<br> Hengyuan Zhao, Xiangtao Kong, Jingwen He, Yu Qiao, Chao Dong<br> Accepted at ECCVW'20<br> :scroll:paper :computer:code

• ESRGAN: Enhanced Super-Resolution Generative Adversarial Networks<br> Xintao Wang, Ke Yu, Shixiang Wu, Jinjin Gu, Yihao Liu, Chao Dong, Yu Qiao, Chen Change Loy<br> Accepted at ECCVW'18<br> :scroll:paper :computer:code

• Recovering Realistic Texture in Image Super-Resolution by Deep Spatial Feature Transform<br> Accepted at CVPR'18<br> Xintao Wang, Ke Yu, Chao Dong, Chen Change Loy<br> :scroll:paper :computer:code

• Accelerating the Super-Resolution Convolutional Neural Network<br> Chao Dong, Chen Change Loy, Xiaoou Tang<br> Accepted at ECCV'16<br> :scroll:paper :computer:code

• Image Super-Resolution Using Deep Convolutional Networks<br> Chao Dong, Chen Change Loy, Kaiming He, Xiaoou Tang<br> Accepted at TPAMI'16<br> :scroll:paper :house:project

License

This project is released under the Apache 2.0 license.

Projects in Open-XSource

• X-Super Resolution: Algorithms in the realm of image super-resolution.
• X-Image Processing: Algorithms in the realm of image restoration and enhancement.
• X-Video Processing: Algorithms for processing videos.
• X-Low level Interpretation: Algorithms for interpreting the principle of neural networks in low-level vision field.
• X-Evaluation and Benchmark: Datasets for training or evaluating state-of-the-art algorithms.