CVPR 2023 超分辨率（super-resolution）方向上接收论文总结

CVPR 2023

官网链接：https://cvpr2023.thecvf.com/
会议时间：2023年6月18日-6月22日，加拿大温哥华
CVPR 2023统计数据：

• 提交：9155篇论文
• 接受：2359篇论文（25.8%的接受率）
• 亮点：235篇论文（占录取论文的10%，占提交论文的2.6%）
• 获奖候选人：12篇论文（占录取论文的0.51%，占提交论文的0.13%）

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. N-Gram in Swin Transformers for Efficient Lightweight Image Super-Resolution
   • Paper: https://arxiv.org/abs/2211.11436
   • Code: https://github.com/rami0205/NGramSwin
   • Keywords: Transformer, Lightweight
2. Perception-Oriented Single Image Super-Resolution using Optimal Objective Estimation
   • Paper: https://arxiv.org/abs/2211.13676
   • Code: https://github.com/seungho-snu/SROOE
3. Activating More Pixels in Image Super-Resolution Transformer
   • Paper: https://arxiv.org/abs/2205.04437
   • Code: https://github.com/XPixelGroup/HAT
   • Keywords: Transformer
4. Burstormer: Burst Image Restoration and Enhancement Transformer
   • Paper: https://arxiv.org/abs/2304.01194
   • Code: http://github.com/akshaydudhane16/Burstormer
   • Keywords: Burst super-resolution
5. Generative Diffusion Prior for Unified Image Restoration and Enhancement
   • Paper: https://arxiv.org/abs/2304.01247
   • Keywords: Unified image recovery
6. Tunable Convolutions with Parametric Multi-Loss Optimization
   • Paper: https://arxiv.org/abs/2304.00898
7. Omni Aggregation Networks for Lightweight Image Super-Resolution
   • Paper: https://arxiv.org/abs/2304.10244
   • Code: https://github.com/Francis0625/Omni-SR
   • Keywords: Lightweight
8. CABM: Content-Aware Bit Mapping for Single Image Super-Resolution Network with Large Input
   • Paper: https://arxiv.org/abs/2304.06454
   • Keywords: Large Input
9. Image Super-Resolution Using T-Tetromino Pixels
   • Paper: https://arxiv.org/abs/2111.09013
10. Spectral Bayesian Uncertainty for Image Super-resolution
    • Paper:
11. Memory-friendly Scalable Super-resolution via Rewinding Lottery Ticket Hypothesis
    • Paper:
    • News: PAMI中心8项研究成果被计算机视觉顶级会议CVPR2023录用
    • Keywords: Lightweight

任意尺度超分

1. Local Implicit Normalizing Flow for Arbitrary-Scale Image Super-Resolution
   • Paper: https://arxiv.org/abs/2303.05156
   • Keywords: Arbitrary-Scale, Flow
2. Super-Resolution Neural Operator
   • Paper: https://arxiv.org/abs/2303.02584
   • Code: https://github.com/2y7c3/Super-Resolution-Neural-Operator
   • Keywords: Arbitrary-Scale
3. OPE-SR: Orthogonal Position Encoding for Designing a Parameter-free Upsampling Module in Arbitrary-scale Image Super-Resolution
   • Paper: https://arxiv.org/abs/2303.01091
   • Keywords: Arbitrary-scale
4. Human Guided Ground-truth Generation for Realistic Image Super-resolution
   • Paper: https://arxiv.org/abs/2303.13069
   • Code: https://github.com/ChrisDud0257/HGGT
   • Keywords: RealSR
5. Cascaded Local Implicit Transformer for Arbitrary-Scale Super-Resolution
   • Paper: https://arxiv.org/abs/2303.16513
   • Code: https://github.com/jaroslaw1007/CLIT
6. Implicit Diffusion Models for Continuous Super-Resolution
   • Paper: https://arxiv.org/abs/2303.16491
   • Code: https://github.com/ree1s/idm
7. CiaoSR: Continuous Implicit Attention-in-Attention Network for Arbitrary-Scale Image Super-Resolution
   • Paper: https://arxiv.org/abs/2212.04362
   • Code: https://github.com/caojiezhang/CiaoSR
   • Keywords: Attention, Implicit
8. Deep Arbitrary-Scale Image Super-Resolution via Scale-Equivariance Pursuit
   • Paper:
   • Code: https://github.com/neuralchen/EQSR

盲超分

1. Better “CMOS” Produces Clearer Images: Learning Space-Variant Blur Estimation for Blind Image Super-Resolution
   • Paper: https://arxiv.org/abs/2304.03542
2. Learning Generative Structure Prior for Blind Text Image Super-resolution
   • Paper: https://arxiv.org/abs/2303.14726
   • Code: https://github.com/csxmli2016/MARCONet

视频超分

1. Learning Spatial-Temporal Implicit Neural Representations for Event-Guided Video Super-Resolution
   • Paper: https://arxiv.org/abs/2303.13767
   • Code: http://github.io/cvpr23/egvsr
   • Keywords: Implicit Neural Representations
2. Towards High-Quality and Efficient Video Super-Resolution via Spatial-Temporal Data Overfitting
   • Paper: https://arxiv.org/abs/2303.08331
   • Code: https://github.com/coulsonlee/STDO-CVPR2023.git
3. Consistent Direct Time-of-Flight Video Depth Super-Resolution
   • Paper: https://arxiv.org/abs/2211.08658
   • Keywords: dToF
4. Compression-Aware Video Super-Resolution
   • Paper:
5. Structured Sparsity Learning for Efficient Video Super-Resolution
   • Paper: https://arxiv.org/abs/2206.07687
   • Code: https://github.com/Zj-BinXia/SSL

特殊场景

1. Learning to Zoom and Unzoom
   • Paper: https://arxiv.org/abs/2303.15390
   • Code: https://tchittesh.github.io/lzu/
   • Keywords: Image Resampling
2. Toward Stable, Interpretable, and Lightweight Hyperspectral Super-resolution
   • Code: https://github.com/WenjinGuo/DAEM
   • Keywords: Hyperspectral
3. OSRT: Omnidirectional Image Super-Resolution with Distortion-aware Transformer
   • Paper: https://arxiv.org/abs/2302.03453
   • Code: https://github.com/Fanghua-Yu/OSRT
   • Keywords: Omnidirectional Image
4. Cross-Guided Optimization of Radiance Fields with Multi-View Image Super-Resolution for High-Resolution Novel View Synthesis
   • Paper:
5. Guided Depth Super-Resolution by Deep Anisotropic Diffusion
   • Paper: https://arxiv.org/abs/2211.11592
   • Code: https://github.com/prs-eth/Diffusion-Super-Resolution
   • Keywords: Depth image, Diffusion
6. CutMIB: Boosting Light Field Super-Resolution via Multi-View Image Blending
   • Paper:
   • Keywords: Light Field
   • Author: http://staff.ustc.edu.cn/~zwxiong/
7. B-spline Texture Coefficients Estimator for Screen Content Image Super-Resolution
   • Paper: https://ipl.dgist.ac.kr/BTC_cvpr23.pdf
   • Code: https://github.com/ByeongHyunPak/btc
   • Keywords: Screen Content Image
8. Spatial-Frequency Mutual Learning for Face Super-Resolution
   • Paper:
   • Keywords: Face
9. Equivalent Transformation and Dual Stream Network Construction for Mobile Image Super-Resolution
   • Paper:
   • Keywords: Mobile Image
10. Zero-Shot Dual-Lens Super-Resolution
    • Paper:
    • Code: https://github.com/XrKang/ZeDuSR
    • Keywords: Zero-Shot, Dual-Lens

总结

从本届接收的论文来看，超分方向上目前主要聚焦于任意尺度超分（ Arbitrary-Scale SR）。

参考资料

1. CVPR2023最新信息及论文下载（Papers/Codes/Project/PaperReading／Demos/直播分享／论文分享会等）
2. Awesome-Super-Resolution
3. CVPR 2023 Accepted Papers