CVPR2022论文速递（2022.3.23）！共17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

欢迎关注：

φ-SfT: Shape-from-Template with a Physics-Based Deformation Model

• 论文/Paper: http://arxiv.org/abs/2203.11938

• 代码/Code: None

Dataset Distillation by Matching Training Trajectories

• 论文/Paper: http://arxiv.org/abs/2203.11932

• 代码/Code: https://github.com/GeorgeCazenavette/mtt-distillation

GradViT: Gradient Inversion of Vision Transformers

• 论文/Paper: http://arxiv.org/abs/2203.11894

• 代码/Code: https://gradvit.github.io/

AP-BSN: Self-Supervised Denoising for Real-World Images via Asymmetric PD and Blind-Spot Network

• 论文/Paper: http://arxiv.org/abs/2203.11799

• 代码/Code: None

Exploring and Evaluating Image Restoration Potential in Dynamic Scenes

• 论文/Paper: http://arxiv.org/abs/2203.11754

• 代码/Code: None

Meta-attention for ViT-backed Continual Learning

• 论文/Paper: http://arxiv.org/abs/2203.11684

• 代码/Code: None

Look for the Change: Learning Object States and State-Modifying Actions from Untrimmed Web Videos

• 论文/Paper: http://arxiv.org/abs/2203.11637

• 代码/Code: https://github.com/soCzech/LookForTheChange

IDEA-Net: Dynamic 3D Point Cloud Interpolation via Deep Embedding Alignment

• 论文/Paper: http://arxiv.org/abs/2203.11590

• 代码/Code: None

Unsupervised Deraining: Where Contrastive Learning Meets Self-similarity

• 论文/Paper: http://arxiv.org/abs/2203.11509

• 代码/Code: None

TransFusion: Robust LiDAR-Camera Fusion for 3D Object Detection with Transformers

• 论文/Paper: http://arxiv.org/abs/2203.11496

• 代码/Code: https://github.com/XuyangBai/TransFusion

Practical Stereo Matching via Cascaded Recurrent Network with Adaptive Correlation

• 论文/Paper: http://arxiv.org/abs/2203.11483

• 代码/Code: None

Mixed Differential Privacy in Computer Vision

• 论文/Paper: http://arxiv.org/abs/2203.11481

• 代码/Code: None

Remember Intentions: Retrospective-Memory-based Trajectory Prediction

• 论文/Paper: http://arxiv.org/abs/2203.11474

• 代码/Code: None

Ray3D: ray-based 3D human pose estimation for monocular absolute 3D localization

• 论文/Paper: http://arxiv.org/abs/2203.11471

• 代码/Code: https://github.com/YxZhxn/Ray3D

Global Matching with Overlapping Attention for Optical Flow Estimation

• 论文/Paper: http://arxiv.org/abs/2203.11335

• 代码/Code: None

NeRFusion: Fusing Radiance Fields for Large-Scale Scene Reconstruction

• 论文/Paper: http://arxiv.org/abs/2203.11283

• 代码/Code: None

Learning from All Vehicles

• 论文/Paper: http://arxiv.org/abs/2203.11934

• 代码/Code: https://github.com/dotchen/LAV