CVPR2022论文速递（2022.6.14）！共9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

大家好, 最近正在优化每周分享的CVPR论文, 目前考虑按照不同类别去分类,方便不同方向的小伙伴挑选自己感兴趣的论文哈

欢迎大家留言其他想法, 合适的话会采纳哈!

Updated on : 14 Jun 2022

total number : 9

Transformers - - 2 篇

Exploring Structure-aware Transformer over Interaction Proposals for Human-Object Interaction Detection

标题：探索与人类相互作用检测的相互作用建议探索结构感知的Transformer

• 论文/Paper: http://arxiv.org/pdf/2206.06291
• 代码/Code: https://github.com/zyong812/STIP

Graph-based Spatial Transformer with Memory Replay for Multi-future Pedestrian Trajectory Prediction

标题：基于图的空间Transformer具有内存重播，用于多未来的行人轨迹预测

• 论文/Paper: http://arxiv.org/pdf/2206.05712
• 代码/Code: https://github.com/Jacobieee/ST-MR.

轨迹预测/Trajectory Prediction - 1 篇

Graph-based Spatial Transformer with Memory Replay for Multi-future Pedestrian Trajectory Prediction

标题：基于图的空间Transformer具有内存重播，用于多未来的行人轨迹预测

• 论文/Paper: http://arxiv.org/pdf/2206.05712
• 代码/Code: https://github.com/Jacobieee/ST-MR.

NeRF - 1 篇

AR-NeRF: Unsupervised Learning of Depth and Defocus Effects from Natural Images with Aperture Rendering Neural Radiance Fields

标题：AR-NeRF：无监督的深度和散焦效应的自然图像带有神经辐射场的效果

• 论文/Paper: http://arxiv.org/pdf/2206.06100
• 代码/Code: None

其他/Other - 6 篇

MLP-3D: A MLP-like 3D Architecture with Grouped Time Mixing

标题：MLP-3D：类似于MLP的3D体系结构，具有分组时间混合

• 论文/Paper: http://arxiv.org/pdf/2206.06292
• 代码/Code: https://github.com/ZhaofanQiu/MLP-3D

Learning Feature Disentanglement and Dynamic Fusion for Recaptured Image Forensic

标题：学习功能分离和动态融合，用于重新捕获图像取证

• 论文/Paper: http://arxiv.org/pdf/2206.06103
• 代码/Code: None

An Evaluation of OCR on Egocentric Data

标题：OCR对以自我为中心数据的评估

• 论文/Paper: http://arxiv.org/pdf/2206.05496
• 代码/Code: None

Precise Affordance Annotation for Egocentric Action Video Datasets

标题：以自我为中心动作视频数据集的精确负担性注释

• 论文/Paper: http://arxiv.org/pdf/2206.05424
• 代码/Code: None

Object Instance Identification in Dynamic Environments

标题：动态环境中的对象实例标识

• 论文/Paper: http://arxiv.org/pdf/2206.05319
• 代码/Code: None

Less Is More: Linear Layers on CLIP Features as Powerful VizWiz Model

标题：少更多：剪辑功能上的线性层作为强大的Vizwiz模型

• 论文/Paper: http://arxiv.org/pdf/2206.05281
• 代码/Code: None