CVPR2022论文速递（2022.4.8）！共18篇

整理：AI算法与图像处理

CVPR2022论文和代码整理：https://github.com/DWCTOD/CVPR2022-Papers-with-Code-Demo

语义分割/Segmentation - 3 篇

Pin the Memory: Learning to Generalize Semantic Segmentation

标题：针内存：学习概括语义细分

• 论文/Paper: http://arxiv.org/pdf/2204.03609
• 代码/Code: None

Coarse-to-Fine Feature Mining for Video Semantic Segmentation

标题：用于视频语义分割的粗对细特征挖掘

• 论文/Paper: http://arxiv.org/pdf/2204.03330
• 代码/Code: https://github.com/guoleisun/vss-cffm

L2G: A Simple Local-to-Global Knowledge Transfer Framework for Weakly Supervised Semantic Segmentation

标题：L2G：一个简单的本地对全球知识转移框架，用于弱监督语义分割

• 论文/Paper: http://arxiv.org/pdf/2204.03206
• 代码/Code: https://github.com/PengtaoJiang/L2G.

GAN - 1 篇

Unsupervised Image-to-Image Translation with Generative Prior

标题：与生成的未经监督的图像到图像转换

• 论文/Paper: http://arxiv.org/pdf/2204.03641
• 代码/Code: https://github.com/williamyang1991/gp-unit

Transformers - - 1 篇

PSTR: End-to-End One-Step Person Search With Transformers

标题：PSTR：结束到最后的一步人与变压器搜索

• 论文/Paper: http://arxiv.org/pdf/2204.03340
• 代码/Code: https://github.com/jialecao001/pstr

对比学习/Contrastive Learning - 1 篇

Unified Contrastive Learning in Image-Text-Label Space

标题：在图像文本标签空间中的统一对比学习

• 论文/Paper: http://arxiv.org/pdf/2204.03610
• 代码/Code: https://github.com/microsoft/unicl

视频插帧/Frame Interpolation - 1 篇

Many-to-many Splatting for Efficient Video Frame Interpolation

标题：有效的视频帧插值多对多分裂

• 论文/Paper: http://arxiv.org/pdf/2204.03513
• 代码/Code: https://github.com/feinanshan/m2m_vfi

其他/Other - 11 篇

Total Variation Optimization Layers for Computer Vision

标题：计算机视觉的总变化优化层

• 论文/Paper: http://arxiv.org/pdf/2204.03643
• 代码/Code: https://github.com/raymondyeh07/tv_layers_for_cv

Pre-train, Self-train, Distill: A simple recipe for Supersizing 3D Reconstruction

标题：火车前，自动列车，蒸馏：一个简单的配方，用于超出3D重建

• 论文/Paper: http://arxiv.org/pdf/2204.03642
• 代码/Code: None

Class-Incremental Learning with Strong Pre-trained Models

标题：Class-Incremental学习与强大的预先训练模型

• 论文/Paper: http://arxiv.org/pdf/2204.03634
• 代码/Code: None

AutoRF: Learning 3D Object Radiance Fields from Single View Observations

标题：autorf：从单视图观察中学习3D对象辐射字段

• 论文/Paper: http://arxiv.org/pdf/2204.03593
• 代码/Code: None

Deep Visual Geo-localization Benchmark

标题：深度视觉地理定位基准

• 论文/Paper: http://arxiv.org/pdf/2204.03444
• 代码/Code: None

Winoground: Probing Vision and Language Models for Visio-Linguistic Compositionality

标题：Winoground：Visio-linguisticsitality的探测视觉和语言模型

• 论文/Paper: http://arxiv.org/pdf/2204.03162
• 代码/Code: None

UIGR: Unified Interactive Garment Retrieval

标题：UIGR：统一互动服装检索

• 论文/Paper: http://arxiv.org/pdf/2204.03111
• 代码/Code: https://github.com/brandonhanx/compfashion

AUV-Net: Learning Aligned UV Maps for Texture Transfer and Synthesis

标题：AUV-NET：学习对齐的UV地图，用于纹理转移和合成

• 论文/Paper: http://arxiv.org/pdf/2204.03105
• 代码/Code: None

Hierarchical Self-supervised Representation Learning for Movie Understanding

标题：电影理解的分层自我监督的代表学习

• 论文/Paper: http://arxiv.org/pdf/2204.03101
• 代码/Code: None

Learning from Untrimmed Videos: Self-Supervised Video Representation Learning with Hierarchical Consistency

标题：从未经监控的视频中学习：自我监督视频表示学习，具有分层一致性

• 论文/Paper: http://arxiv.org/pdf/2204.03017
• 代码/Code: None

Multi-Scale Memory-Based Video Deblurring

标题：基于多尺度内存的视频去纹理

• 论文/Paper: http://arxiv.org/pdf/2204.02977
• 代码/Code: https://github.com/jibo27/memdeblur