CVPR2022论文速递（2022.4.18）！共9篇！多篇成果代码已开源！

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

欢迎关注公众号 AI算法与图像处理，获取更多干货：

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! 求个三连支持一波哈

Updated on : 18 Apr 2022 total number : 9 CVPR2022 Oral - 1 篇

Learning Multi-View Aggregation In the Wild for Large-Scale 3D Semantic Segmentation  (Oral)

标题：在野外学习多视图聚集，用于大规模3D语义分割（ORAL）

• 论文/Paper: http://arxiv.org/pdf/2204.07548

• 代码/Code: https://github.com/drprojects/DeepViewAgg

目标检测 / Object Detection - 1 篇

Dense Learning based Semi-Supervised Object Detection

标题：基于稠密的学习半监督对象检测

• 论文/Paper: http://arxiv.org/pdf/2204.07300

• 代码/Code: https://github.com/chenbinghui1/DSL

语义分割/Segmentation - 1 篇

Learning Multi-View Aggregation In the Wild for Large-Scale 3D Semantic Segmentation  (Oral)

标题：在野外学习多视图聚集，用于大规模3D语义分割（ORAL）

• 论文/Paper: http://arxiv.org/pdf/2204.07548

• 代码/Code: https://github.com/drprojects/DeepViewAgg

车道线检测/Lane Detection - 1 篇

A Keypoint-based Global Association Network for Lane Detection

标题：基于关键点的全局关联网络，用于车道检测

• 论文/Paper: http://arxiv.org/pdf/2204.07335

• 代码/Code: https://github.com/Wolfwjs/GANet.

其他/Other - 6 篇

Patch-wise Contrastive Style Learning for Instagram Filter Removal

标题：用于Instagram Filter移除的Patch-wise的对比风格学习

• 论文/Paper: http://arxiv.org/pdf/2204.07486

• 代码/Code: None

Guiding Attention using Partial-Order Relationships for Image Captioning

标题：使用Partial-Order关系引导attention图像描述

• 论文/Paper: http://arxiv.org/pdf/2204.07476

• 代码/Code: None

MetaSets: Meta-Learning on Point Sets for Generalizable Representations

标题：MetaSets：元学习在通用表示的点集合

• 论文/Paper: http://arxiv.org/pdf/2204.07311

• 代码/Code: None

Pushing the Limits of Simple Pipelines for Few-Shot Learning: External Data and Fine-Tuning Make a Difference

标题：推动简单管道用于Few-Shot学习的极限：外部数据和微调会产生差异

• 论文/Paper: http://arxiv.org/pdf/2204.07305

• 代码/Code: None

Imposing Consistency for Optical Flow Estimation

标题：对光流估计的施加一致性

• 论文/Paper: http://arxiv.org/pdf/2204.07262

• 代码/Code: None

Measuring Compositional Consistency for Video Question Answering

标题：测量视频问题问答的组成一致性

• 论文/Paper: http://arxiv.org/pdf/2204.07190

• 代码/Code: None