CVPR2022论文速递（2022.4.15）！共16篇！内含2篇Oral！

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

Updated on : 15 Apr 2022

total number : 16

CVPR2022 Oral - 2 篇

Joint Forecasting of Panoptic Segmentations with Difference Attention (Oral)

标题：差异关注联合预测全景分割

• 论文/Paper: http://arxiv.org/pdf/2204.07157
• 代码/Code: None

Deformable Sprites for Unsupervised Video Decomposition (Oral)

标题：无监督视频分解的可变形Sprites

• 论文/Paper: http://arxiv.org/pdf/2204.07151
• 代码/Code: None

目标跟踪/Object Tracking - 2 篇

BEHAVE: Dataset and Method for Tracking Human Object Interactions

标题：BEHAVE：数据集和用于跟踪人类对象交互的方法

• 论文/Paper: http://arxiv.org/pdf/2204.06950
• 代码/Code: None

SoccerNet-Tracking: Multiple Object Tracking Dataset and Benchmark in Soccer Videos

标题：SoccerNet跟踪：多个对象跟踪数据集和足球视频中的基准

• 论文/Paper: http://arxiv.org/pdf/2204.06918
• 代码/Code: None

语义分割/Segmentation - 2 篇

Joint Forecasting of Panoptic Segmentations with Difference Attention (Oral)

标题：差异关注联合预测全景分割

• 论文/Paper: http://arxiv.org/pdf/2204.07157
• 代码/Code: None

Cross-Image Relational Knowledge Distillation for Semantic Segmentation

标题：语义分割的交叉图像关系知识蒸馏

• 论文/Paper: http://arxiv.org/pdf/2204.06986
• 代码/Code: https://github.com/winycg/cirkd

超分/Super-Resolution - 1 篇

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

标题：来回看：视频超分辨率与明确的时间差异建模

• 论文/Paper: http://arxiv.org/pdf/2204.07114
• 代码/Code: None

Transformers - - 2 篇

MiniViT: Compressing Vision Transformers with Weight Multiplexing

标题：MinioIVIT：压缩具有重量复用的视觉Transformers

• 论文/Paper: http://arxiv.org/pdf/2204.07154
• 代码/Code: https://github.com/microsoft/cream

ViTOL: Vision Transformer for Weakly Supervised Object Localization

标题：Vitol：视觉Transformers，用于弱监督对象本地化

• 论文/Paper: http://arxiv.org/pdf/2204.06772
• 代码/Code: https://github.com/Saurav-31/ViTOL

行人重识别/Person Re-Identification - 2 篇

Implicit Sample Extension for Unsupervised Person Re-Identification

标题：无监督行人重识别的隐式示例扩展

• 论文/Paper: http://arxiv.org/pdf/2204.06892
• 代码/Code: https://github.com/PaddlePaddle/PaddleClas

Clothes-Changing Person Re-identification with RGB Modality Only

标题：改变的人只用RGB模态行人重识别

• 论文/Paper: http://arxiv.org/pdf/2204.06890
• 代码/Code: https://github.com/guxinqian/Simple-CCReID.

其他/Other - 6 篇

What's in your hands? 3D Reconstruction of Generic Objects in Hands

标题：你手中的是什么？三维重建在手中的通用物体

• 论文/Paper: http://arxiv.org/pdf/2204.07153
• 代码/Code: None

GIFS: Neural Implicit Function for General Shape Representation

标题：GIFS：一般形状表示的神经隐式功能

• 论文/Paper: http://arxiv.org/pdf/2204.07126
• 代码/Code: None

The multi-modal universe of fast-fashion: the Visuelle 2.0 benchmark

标题：fast-fashion的多模态宇宙：VISUELLE 2.0基准

• 论文/Paper: http://arxiv.org/pdf/2204.06972
• 代码/Code: None

Semi-Supervised Training to Improve Player and Ball Detection in Soccer

标题：半监督训练，以改善足球运动员和球侦查

• 论文/Paper: http://arxiv.org/pdf/2204.06859
• 代码/Code: https://github.com/rvandeghen/SST

Pyramidal Attention for Saliency Detection

标题：显着性检测的金字塔Attention

• 论文/Paper: http://arxiv.org/pdf/2204.06788
• 代码/Code: https://github.com/tanveer-hussain/EfficientSOD2

OccAM's Laser: Occlusion-based Attribution Maps for 3D Object Detectors on LiDAR Data

标题：Incam的激光：LIDAR数据上的3D对象探测器的基于遮挡的归属映射

• 论文/Paper: http://arxiv.org/pdf/2204.06577
• 代码/Code: https://github.com/dschinagl/occam