CVPR 2021
CVPR2021官网:http://cvpr2021.thecvf.com
开会时间:2021年6月19日-6月25日(线上)
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
- Cross-MPI: Cross-scale Stereo for Image Super-Resolution using Multiplane Images
- Paper:https://arxiv.org/abs/2011.14631
- Code:
- Homepage:http://www.liuyebin.com/crossMPI/crossMPI.html
- Analysis:CVPR 2021,Cross-MPI以底层场景结构为线索的端到端网络,在大分辨率(x8)差距下也可完成高保真的超分辨率
- Data-Free Knowledge Distillation For Image Super-Resolution
- Keywords: DAFL算法的SR版本,华为诺亚
- LAU-Net: Latitude Adaptive Upscaling Network for Omnidirectional Image Super-resolution
- Learning Continuous Image Representation with Local Implicit Image Function
- Paper:https://arxiv.org/abs/2012.09161
- Code:https://github.com/yinboc/liif
- Homepage:https://yinboc.github.io/liif/
- Analysis:真正的无极放大!30x 插值效果惊艳,英伟达等开源 LIIF:巧妙的图像超分新思路
- Robust Reference-based Super-Resolution via C²-Matching
- GLEAN: Generative Latent Bank for Large-Factor Image Super-Resolution
- Paper:https://ckkelvinchan.github.io/papers/glean.pdf
- Code:
- Homepage:https://ckkelvinchan.github.io/projects/GLEAN/
- Keywords:商汤,GAN
- Analysis:Generative Latent Bank for Large-Factor Image Super-Resolution
- MASA-SR: Matching Acceleration and Spatial Adaptation for Reference-Based Image Super-Resolution
- Interpreting Super-Resolution Networks with Local Attribution Maps
- SRWarp: Generalized Image Super-Resolution under Arbitrary Transformation
轻量图像超分
- AdderSR: Towards Energy Efficient Image Super-Resolution (oral)
- Paper:https://arxiv.org/abs/2009.08891
- Code:
- Keywords:加法神经网络(CVPR2020 Oral)(AdderNet)的工作。
- ClassSR: A General Framework to Accelerate Super-Resolution Networks by Data Characteristic
- Paper:https://arxiv.org/abs/2103.04039
- Code:https://github.com/Xiangtaokong/ClassSR
- Keywords:classification,lightweight,董超老师组的工作
- Analysis:CVPR2021|超分性能不降低,计算量降低50%,董超等人提出加速图像超分的ClassSR
- Exploring Sparsity in Image Super-Resolution for Efficient Inference
- Paper:https://arxiv.org/abs/2006.09603
- Code:https://github.com/LongguangWang/SMSR
- Keywords:计算资源
- Analysis:CVPR2021 | 性能不变,计算量减少41%,国防科大提出加速图像超分高效推理的SMSR
无监督
- Unsupervised Degradation Representation Learning for Blind Super-Resolution
- Flow-based Kernel Prior with Application to Blind Super-Resolution
- Paper:https://arxiv.org/abs/2103.15977
- Code:https://github.com/JingyunLiang/FKP
- Keywords:Blind SR, flow-based kernel prio
视频超分
- BasicVSR: The Search for Essential Components in Video Super-Resolution and Beyond
- Paper:https://arxiv.org/abs/2012.02181
- Code:https://github.com/ckkelvinchan/BasicVSR-IconVSR
- Homepage:https://ckkelvinchan.github.io/projects/BasicVSR/
- Keywords:NTIRE2021 冠军,商汤
- Analysis:新视频超分算法来了:CVPR 2021 & NTIRE2021 冠军
- Video Rescaling Networks with Joint Optimization Strategies for Downscaling and Upscaling
- Paper:https://arxiv.org/abs/2103.14858
- Code:https://github.com/ding3820/MIMO-VRN
- Homepage:https://ding3820.github.io/MIMO-VRN/
- Temporal Modulation Network for Controllable Space-Time Video Super-Resolution
- Code:https://github.com/CS-GangXu/TMNet 404
- Researcher:https://csjunxu.github.io/
- Keywords:腾讯优图,时间建模网络
其他场景超分
- Learning Scene Structure Guidance via Cross-Task Knowledge Transfer for Single Depth Super-Resolution
- Paper:http://faculty.dlut.edu.cn/_resources/group1/M00/00/60/ynZMhGBQGYqAesPpADBWp0S7dnM029.pdf
- Code:https://github.com/Sunbaoli/dsr-distillation
- Homepage:http://faculty.dlut.edu.cn/yexinchen/zh_CN/zdylm/1123985/list/index.htm
- Keywords:深度图像超分
- Analysis:CVPR 2021 | 基于跨任务场景结构知识迁移的单张深度图像超分辨率方法
交叉
- Generic Perceptual Loss for Modeling Structured Output Dependencies
- Paper:https://arxiv.org/abs/2103.10571
- Keywords:work at semantic segmentation, depthestimation and instance segmentation
- Analysis:你的感知损失可能用错了,沈春华团队提出随机权值广义感知损失
AAAI 2021
官网链接:https://aaai.org/Conferences/AAAI-21/
AAAI 的英文全称是 Association for the Advance of Artificial Intelligence——美国人工智能协会。该协会是人工智能领域的主要学术组织之一,其主办的年会也是人工智能领域的国际顶级会议。在中国计算机学会的国际学术会议排名以及清华大学新发布的计算机科学推荐学术会议和期刊列表中,AAAI 均被列为人工智能领域的 A 类顶级会议。
AAAI 2021一共收到9034篇论文提交,其中有效审稿有7911篇,最终录取的数量为1692篇,接收率为21.4%。本届会议于2月2日-2月9日全程线上举办。
视频超分
- Understanding Deformable Alignment in Video Super-Resolution
- Paper:https://arxiv.org/abs/2009.07265
- Analysis:AAAI 2021 | 视频超分中的可变形对齐解读
- Large Motion Video Super-Resolution with Dual Subnet and Multi-Stage Communicated Upsampling
- Paper:https://arxiv.org/abs/2103.11744
- Keywords:VSR, Dual Subnet, Multi-stage Communicated Up-sampling