ICCV 2021超分辨率（super-resolution）方向上接收论文总结

ICCV 2021

• ICCV2021官网：https://iccv2021.thecvf.com/
• ICCV2021完整论文列表：https://openaccess.thecvf.com/ICCV2021
• 开会时间：2021年10月11日-10月17日
• Demo 搬运：https://space.bilibili.com/288489574

Super-Resolution

- 图像超分辨率

1. Attention-Based Multi-Reference Learning for Image Super-Resolution

• 论文：http://arxiv.org/abs/2108.13697
• 代码：https://marcopesavento.github.io/AMRSR
• 描述：RefSR

2. Learning a Single Network for Scale-Arbitrary Super-Resolution

• 论文：http://arxiv.org/abs/2004.03791
• 代码：https://github.com/LongguangWang/ArbSR
• 描述：scale-arbitrary 任意缩放

3. Achieving On-Mobile Real-Time Super-Resolution With Neural Architecture and Pruning Search

• 论文：http://arxiv.org/abs/2108.08910
• 代码：
• 描述：real-time, NAS

4. Designing a Practical Degradation Model for Deep Blind Image Super-Resolution

• 论文：http://arxiv.org/abs/2103.14006
• 代码：
• 描述：Blind SR

5. Event Stream Super-Resolution via Spatiotemporal Constraint Learning

• 论文：openaccess.thecvf.com
• 代码：
• 描述：event stream SR

6. Dynamic High-Pass Filtering and Multi-Spectral Attention for Image Super-Resolution

• 论文：openaccess.thecvf.com
• 代码：
• 描述：attention

7. Context Reasoning Attention Network for Image Super-Resolution

• 论文：openaccess.thecvf.com
• 代码：
• 描述：attention

8. EvIntSR-Net: Event Guided Multiple Latent Frames Reconstruction and Super-Resolution

• 论文：openaccess.thecvf.com
• 代码：
• 描述：EventSR
• Demo：AI视频修复！基于Event引导的视频帧重建和超分辨率 ICCV2021

9. Mutual Affine Network for Spatially Variant Kernel Estimation in Blind Image Super-Resolution

• 论文：http://arxiv.org/abs/2108.05302
• 代码：https://github.com/JingyunLiang/MANet
• 描述：blind SR

10. Unsupervised Real-World Super-Resolution: A Domain Adaptation Perspective

• 论文：openaccess.thecvf.com
• 代码：
• 描述：unpaired realSR

11. Hierarchical Conditional Flow: A Unified Framework for Image Super-Resolution and Image Rescaling

• 论文：http://arxiv.org/abs/2108.05301
• 代码：https://github.com/JingyunLiang/HCFlow
• 描述：SR/Rescale

12. Learning Frequency-aware Dynamic Network for Efficient Super-Resolution

• 论文：http://arxiv.org/abs/2103.08357
• 代码：
• 描述：Efficient SR

13. Dual-Camera Super-Resolution With Aligned Attention Modules

• 论文：openaccess.thecvf.com
• 代码：
• 描述：dual camera SR

14. Manifold Matching via Deep Metric Learning for Generative Modeling

• 论文：https://arxiv.org/abs/2106.10777
• 代码：https://github.com/dzld00/pytorch-manifold-matching.git
• 描述：Manifold Matching

- Burst SR

1. Deep Reparametrization of Multi-Frame Super-Resolution and Denoising

• 论文：http://arxiv.org/abs/2108.08286
• 代码：https://github.com/goutamgmb/deep-rep
• 描述：Multi-Frame

2. Lucas-Kanade Reloaded: End-to-End Super-Resolution From Raw Image Bursts

• 论文：http://arxiv.org/abs/2104.06191
• 代码：https://github.com/bruno-31/lkburst
• 描述：Raw Image Bursts SR

- 视频超分辨率

1. Omniscient Video Super-Resolution

• 论文：http://arxiv.org/abs/2004.03791
• 代码：
• 描述：VSR

2. COMISR: Compression-Informed Video Super-Resolution

• 论文：http://arxiv.org/abs/2105.01237
• 代码：https://github.com/google-research/google-research/tree/master/comisr
• 描述：VSR

3. Deep Blind Video Super-Resolution

• 论文：openaccess.thecvf.com
• 代码：
• 描述：Blind VSR

4. Real-World Video Super-Resolution: A Benchmark Dataset and a Decomposition Based Learning Scheme

• 论文：openaccess.thecvf.com
• 代码：https://github.com/IanYeung/RealVSR
• 描述：RealVSR dataset

5. Efficient Video Compression via Content-Adaptive Super-Resolution

• 论文：http://arxiv.org/abs/2104.02322
• 代码：
• 描述：Video Compression

6. Overfitting the Data: Compact Neural Video Delivery via Content-aware Feature Modulation

• 论文：https://arxiv.org/abs/2108.08202
• 代码：https://github.com/Neural-video-delivery/CaFM-Pytorch-ICCV2021
• 描述：Video Delivery

- 其他

1. Fourier Space Losses for Efficient Perceptual Image Super-Resolution

• 论文：http://arxiv.org/abs/2106.00783
• 代码：
• 描述：loss

2. Benchmarking Ultra-High-Definition Image Super-Resolution

• 论文：openaccess.thecvf.com
• 代码：
• 描述：UHD SR dataset, 4K, 8K,

参考资料

1. ChaofWang / Awesome-Super-Resolution Public
2. ICCV2021 图像超分汇总持续更新中…
3. ICCV 2021 open access
4. Awesome-ICCV2021-Low-Level-Vision