CVPR2022论文速递（2022.4.7）！共17篇，新增标题翻译

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

目标检测 / Object Detection - 1 篇

Towards Robust Adaptive Object Detection under Noisy Annotations

标题：在嘈杂的注释下朝着强大的自适应对象检测

• 论文/Paper: http://arxiv.org/pdf/2204.02620
• 代码/Code: None

语义分割/Segmentation - 2 篇

FocalClick: Towards Practical Interactive Image Segmentation

标题：FOCALCLICK：走向实际交互式图像分割

• 论文/Paper: http://arxiv.org/pdf/2204.02574
• 代码/Code: https://github.com/XavierCHEN34/ClickSEG

Modeling Motion with Multi-Modal Features for Text-Based Video Segmentation

标题：基于文本的视频分割的多模态特征建模运动

• 论文/Paper: http://arxiv.org/pdf/2204.02547
• 代码/Code: None

其他/Other - 15 篇

Temporal Alignment Networks for Long-term Video

标题：用于长期视频的时间对准网络

• 论文/Paper: http://arxiv.org/pdf/2204.02968
• 代码/Code: None

"The Pedestrian next to the Lamppost" Adaptive Object Graphs for Better Instantaneous Mapping

标题：“灯柱旁边的行人”自适应对象图，用于更好的瞬时映射

• 论文/Paper: http://arxiv.org/pdf/2204.02944
• 代码/Code: None

An Empirical Study of End-to-End Temporal Action Detection

标题：端到端时间作用检测的实证研究

• 论文/Paper: http://arxiv.org/pdf/2204.02932
• 代码/Code: https://github.com/xlliu7/E2E-TAD

Masking Adversarial Damage: Finding Adversarial Saliency for Robust and Sparse Network

标题：掩盖对抗性损伤：寻找强大和稀疏网络的对抗显着性

• 论文/Paper: http://arxiv.org/pdf/2204.02738
• 代码/Code: None

Aesthetic Text Logo Synthesis via Content-aware Layout Inferring

标题：通过内容感知布局推断的美学文本徽标综合

• 论文/Paper: http://arxiv.org/pdf/2204.02701
• 代码/Code: https://github.com/yizhiwang96/TextLogoLayout

Cloning Outfits from Real-World Images to 3D Characters for Generalizable Person Re-Identification

标题：克隆从真实世界的图像到3D字符的克隆服装，以便重新识别

• 论文/Paper: http://arxiv.org/pdf/2204.02611
• 代码/Code: https://github.com/Yanan-Wang-cs/ClonedPerson

Learning to Anticipate Future with Dynamic Context Removal

标题：使用动态背景删除学习预期未来

• 论文/Paper: http://arxiv.org/pdf/2204.02587
• 代码/Code: https://github.com/AllenXuuu/DCR.

SqueezeNeRF: Further factorized FastNeRF for memory-efficient inference

标题：Squeezenerf：进一步分解Fastnerf，用于记忆高效推论

• 论文/Paper: http://arxiv.org/pdf/2204.02585
• 代码/Code: None

Gait Recognition in the Wild with Dense 3D Representations and A Benchmark

标题：步态在野外的崇拜与密集的3D表示和基准

• 论文/Paper: http://arxiv.org/pdf/2204.02569
• 代码/Code: None

MixFormer: Mixing Features across Windows and Dimensions

标题：混合器：跨窗口和尺寸的混合功能

• 论文/Paper: http://arxiv.org/pdf/2204.02557
• 代码/Code: https://github.com/PaddlePaddle/PaddleClas

RODD: A Self-Supervised Approach for Robust Out-of-Distribution Detection

标题：Rodd：一种自我监督的方法，可实现稳健的分配检测

• 论文/Paper: http://arxiv.org/pdf/2204.02553
• 代码/Code: None

Adversarial Robustness through the Lens of Convolutional Filters

标题：通过卷积滤镜的镜片对抗鲁棒性

• 论文/Paper: http://arxiv.org/pdf/2204.02481
• 代码/Code: https://github.com/paulgavrikov/cvpr22w_RobustnessThroughTheLens

Towards An End-to-End Framework for Flow-Guided Video Inpainting

标题：走向流动导向视频染色的端到端框架

• 论文/Paper: http://arxiv.org/pdf/2204.02663
• 代码/Code: https://github.com/MCG-NKU/E2FGVI.

Learning Optimal K-space Acquisition and Reconstruction using Physics-Informed Neural Networks

标题：学习物理信息神经网络的最佳k空间采集与重建

• 论文/Paper: http://arxiv.org/pdf/2204.02480
• 代码/Code: None