CVPR2022论文速递（2022.4.13）！共17篇！GAN/transformer/多模态等

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 : 13 Apr 2022

total number : 17

分类 / Classification - 1 篇

Regression or Classification? Reflection on BP prediction from PPG data using Deep Neural Networks in the scope of practical applications

标题：回归或分类？基于实际应用范围的深神经网络对PPG数据对BP预测的反射

• 论文/Paper: http://arxiv.org/pdf/2204.05605
• 代码/Code: None

目标检测 / Object Detection - 2 篇

Towards Open-Set Object Detection and Discovery

标题：走向开放式对象检测和发现

• 论文/Paper: http://arxiv.org/pdf/2204.05604
• 代码/Code: None

DAIR-V2X: A Large-Scale Dataset for Vehicle-Infrastructure Cooperative 3D Object Detection

标题：Dair-V2X：用于车辆 - 基础设施协同合作3D对象检测的大规模数据集

• 论文/Paper: http://arxiv.org/pdf/2204.05575
• 代码/Code: https://github.com/AIR-THU/DAIR-V2X.

语义分割/Segmentation - 3 篇

NightLab: A Dual-level Architecture with Hardness Detection for Segmentation at Night

标题：NightLab：一种双层架构，具有夜间分割的硬度检测

• 论文/Paper: http://arxiv.org/pdf/2204.05538
• 代码/Code: None

TopFormer: Token Pyramid Transformer for Mobile Semantic Segmentation

标题：TOPFORMER：移动语义细分的令牌金字塔变压器

• 论文/Paper: http://arxiv.org/pdf/2204.05525
• 代码/Code: https://github.com/hustvl/TopFormer

Panoptic, Instance and Semantic Relations: A Relational Context Encoder to Enhance Panoptic Segmentation

标题：Panoptic，实例和语义关系：关系上下文编码器，以增强Panoptic分段

• 论文/Paper: http://arxiv.org/pdf/2204.05370
• 代码/Code: None

GAN - 1 篇

medXGAN: Visual Explanations for Medical Classifiers through a Generative Latent Space

标题：Medxgan：通过生成潜空间的医疗分类器的视觉解释

• 论文/Paper: http://arxiv.org/pdf/2204.05376
• 代码/Code: ions.

Transformers - - 2 篇

TopFormer: Token Pyramid Transformer for Mobile Semantic Segmentation

标题：TOPFORMER：移动语义细分的令牌金字塔Transformer

• 论文/Paper: http://arxiv.org/pdf/2204.05525
• 代码/Code: https://github.com/hustvl/TopFormer

Are Multimodal Transformers Robust to Missing Modality?

标题：多模式Transformer是否强大地缺少模态？

• 论文/Paper: http://arxiv.org/pdf/2204.05454
• 代码/Code: None

多模态 / Multimodal - 2 篇

Probabilistic Compositional Embeddings for Multimodal Image Retrieval

标题：多式化图像检索的概率组成嵌入

• 论文/Paper: http://arxiv.org/pdf/2204.05845
• 代码/Code: https://github.com/andreineculai/MPC.

Are Multimodal Transformers Robust to Missing Modality?

标题：多模式变压器是否强大地缺少模态？

• 论文/Paper: http://arxiv.org/pdf/2204.05454
• 代码/Code: None

点云/Point Clouds - 1 篇

3DeformRS: Certifying Spatial Deformations on Point Clouds

标题：3DEFORMRS：在点云上认证空间变形

• 论文/Paper: http://arxiv.org/pdf/2204.05687
• 代码/Code: None

检索/Image Retrieval - 1 篇

Probabilistic Compositional Embeddings for Multimodal Image Retrieval

标题：多式化图像检索的概率组成嵌入

• 论文/Paper: http://arxiv.org/pdf/2204.05845
• 代码/Code: https://github.com/andreineculai/MPC.

其他/Other - 7 篇

Arch-Graph: Acyclic Architecture Relation Predictor for Task-Transferable Neural Architecture Search

标题：arch-traph：任务可转让神经架构的非循环架构关系预测器

• 论文/Paper: http://arxiv.org/pdf/2204.05941
• 代码/Code: None

Continual Predictive Learning from Videos

标题：来自视频的持续预测学习

• 论文/Paper: http://arxiv.org/pdf/2204.05624
• 代码/Code: https://github.com/jc043/CPL

HyperDet3D: Learning a Scene-conditioned 3D Object Detector

标题：HyperDET3D：学习场景调节的3D对象探测器

• 论文/Paper: http://arxiv.org/pdf/2204.05599
• 代码/Code: None

Open-set Text Recognition via Character-Context Decoupling

标题：通过字符上下文解耦的开放式文本识别

• 论文/Paper: http://arxiv.org/pdf/2204.05535
• 代码/Code: None

Few-shot Learning with Noisy Labels

标题：嘈杂的标签很少拍摄

• 论文/Paper: http://arxiv.org/pdf/2204.05494
• 代码/Code: None

Out-Of-Distribution Detection In Unsupervised Continual Learning

标题：无监督持续学习的分销检测

• 论文/Paper: http://arxiv.org/pdf/2204.05462
• 代码/Code: None

Generalizing Adversarial Explanations with Grad-CAM

标题：用Grad-Cam推广对抗性解释

• 论文/Paper: http://arxiv.org/pdf/2204.05427
• 代码/Code: None