CVPR 2022 | AAAI2022 | WACV 2022 超分辨率（super-resolution）论文汇总

1. 

作者：yyywxk，https://zhuanlan.zhihu.com/p/500474407

编辑：AI算法与图像处理 （已授权）

文章导读

 

超分辨今年来的研究成果越来越惊艳，特别是在特定场景下的超分，例如人脸超分，以及动漫超分等，相关技术分享总是在b站爆火，今天的文章汇总CVPR2022/AAAI2022/WACV2022上的超分工作！

CVPR2022

官网链接：http://CVPR2022.thecvf.com 会议时间：2021年6月19日-6月24日

现将超分辨率方向上接收的论文汇总如下，遗漏之处还请大家斧正.

图像超分

1. Local Texture Estimator for Implicit Representation Function

• Paper: https://ipl.dgist.ac.kr/LTE_cvpr.pdf

• Code: https://github.com/jaewon-lee-b/lte

2. Reflash Dropout in Image Super-Resolution

• Paper: https://arxiv.org/abs/2112.12089

3. Towards Bidirectional Arbitrary Image Rescaling: Joint Optimization and Cycle Idempotence

• Paper: https://arxiv.org/abs/2203.00911

盲超分

1. Deep Constrained Least Squares for Blind Image Super-Resolution

• Paper: https://arxiv.org/abs/2202.07508

• Code: https://github.com/megvii-research/DCLS-SR

2. Learning the Degradation Distribution for Blind Image Super-Resolution

• Paper: Learning the Degradation Distribution for Blind Image Super-Resolution

• Code: https://github.com/greatlog/UnpairedSR

3. Blind Image Super-resolution with Elaborate Degradation Modeling on Noise and Kernel

• Paper:https://arxiv.org/abs/2107.00986

• Code: https://github.com/zsyOAOA/BSRDM

视频超分

1. Look Back and Forth: Video Super-Resolution with Explicit Temporal Difference Modeling

• Paper: https://arxiv.org/abs/2204.07114

2. RSTT: Real-time Spatial Temporal Transformer for Space-Time Video Super-Resolution

• Paper:https://arxiv.org/abs/2203.14186

• Code: https://github.com/llmpass/RSTT

• Keywords: Transformer

3. Reference-based Video Super-Resolution Using Multi-Camera Video Triplets

• Paper: https://arxiv.org/abs/2203.14537

• Code: https://github.com/codeslake/RefVSR

4. Learning Trajectory-Aware Transformer for Video Super-Resolution

• Paper: https://arxiv.org/abs/2204.04216

• Code: https://github.com/researchmm/TTVSR

特殊场景

1. A Text Attention Network for Spatial Deformation Robust Scene Text Image Super-resolution

• Paper: https://arxiv.org/abs/2203.09388

• Code: https://github.com/mjq11302010044/TATT

• Keywords: Scene Text

2. Details or Artifacts: A Locally Discriminative Learning Approach to Realistic Image Super-Resolution

• Paper: https://arxiv.org/abs/2203.09195

• Code: https://github.com/csjliang/LDL

• Keywords: Realistic

AAAI2022

AAAI是人工智能领域中历史最悠久、涵盖内容最广泛的国际顶级学术会议之一，也是中国计算机学会(CCF)推荐的人工智能领域A级国际学术会议之一。据了解，本届会议收到来自全球的9215篇投稿论文，而接受率仅为15%，创历史新低。

官方网址：https://aaai.org/Conferences/AAAI-22/

图像超分

1. Efficient Non-Local Contrastive Attention for Image Super-Resolution

• Paper: https://arxiv.org/abs/2201.03794

• Code: https://github.com/Zj-BinXia/ENLCA

• Keywords: Attention

2. Best-Buddy GANs for Highly Detailed Image Super-Resolution

• Paper: arxiv.org/abs/2103.1529

• Code: github.com/Jia-Research

• Keywords: GAN

特殊场景

1. Text Gestalt: Stroke-Aware Scene Text Image Super-Resolution
   

• Paper: https://arxiv.org/abs/2112.08171

• Code: https://github.com/FudanVI/FudanOCR/tree/main/text-gestalt

• Keywords: Scene Text

2. Detail-Preserving Transformer for Light Field Image Super-Resolution

• Paper: https://arxiv.org/abs/2201.00346

• Code: https://github.com/BITszwang/DPT

• Keywords: Light Field

3. Coarse-to-Fine Embedded PatchMatch and Multi-Scale Dynamic Aggregation for Reference-Based Super-Resolution

• Paper: https://arxiv.org/abs/2201.04358

• Code: https://github.com/Zj-BinXia/AMSA

• Keywords: Reference-Based

4. Feature Distillation Interaction Weighting Network for Lightweight Image Super-Resolution

• Paper: https://arxiv.org/abs/2112.08655

• Code: https://github.com/IVIPLab/FDIWN

• Keywords: Lightweight

5. SCSNet: An Efficient Paradigm for Learning Simultaneously Image Colorization and Super-Resolution

• Paper: https://arxiv.org/abs/2201.04364

• Keywords: Multi-task

CVPR2022

WACV是IEEE Winter Conference on Applications of Computer Vision缩写，根据谷歌学术显示在 Computer Vision & Pattern Recognition 领域， WACV是排在CVPR \ ICCV \ ECCV \ BMVC 之后的较高质量的国际会议。

官方网址：https://openaccess.thecvf.com/WACV2022

图像超分

1. DAQ: Channel-Wise Distribution-Aware Quantization for Deep Image Super-Resolution Networks
   

• Paper: https://arxiv.org/abs/2012.11230

2. edge-SR: Super-Resolution For The Masses

• Paper: https://arxiv.org/abs/2108.10335

• Code: https://github.com/pnavarre/eSR

视频超分

1. MEGAN: Memory Enhanced Graph Attention Network for Space-Time Video Super-Resolution
   

• Paper: https://arxiv.org/abs/2110.15327

2. Normalizing Flow as a Flexible Fidelity Objective for Photo-Realistic Super-resolution

• Paper: https://arxiv.org/abs/2111.03649

• Code: https://github.com/andreas128/AdFlow?1

• Keywords: Photo-Realistic, Flow

特殊场景

1. Hyperspectral Image Super-Resolution With RGB Image Super-Resolution as an Auxiliary Task
   

• Paper: https://arxiv.org/abs/2101.07589

• Code: https://github.com/kli8996/HSISR

2. MoESR: Blind Super-Resolution Using Kernel-Aware Mixture of Experts

• Paper: https://openaccess.thecvf.com/content/WACV2022/papers/Emad_MoESR_Blind_Super-Resolution_Using_Kernel-Aware_Mixture_of_Experts_WACV_2022_paper.pdf

• Keywords: Blind

3. Multi-Dimensional Dynamic Model Compression for Efficient Image Super-Resolution

• Paper: https://openaccess.thecvf.com/content/WACV2022/papers/Hou_Multi-Dimensional_Dynamic_Model_Compression_for_Efficient_Image_Super-Resolution_WACV_2022_paper.pdf

• Keywords: Model Compression

参考资料

1. https://zhuanlan.zhihu.com/p/478286484

2. https://github.com/ChaofWang/Awesome-Super-Resolution#2022

3. https://github.com/extreme-assistant/CVPR2022-Paper-Code-Interpretation

4.