CVPR2022论文速递（2022.4.22）！共15篇！Transformer+多模态用于食品检索

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 : 22 Apr 2022

total number : 15

超分/Super-Resolution - 2 篇

A New Dataset and Transformer for Stereoscopic Video Super-Resolution

标题：用于立体视频超分辨率的新数据集和Transformer

• 论文/Paper: http://arxiv.org/pdf/2204.10039
• 代码/Code: https://github.com/H-deep/Trans-SVSR/

FS-NCSR: Increasing Diversity of the Super-Resolution Space via Frequency Separation and Noise-Conditioned Normalizing Flow

标题：FS-NCSR：通过频率分离和噪声调节标准化流量增加超分辨率空间的多样性

• 论文/Paper: http://arxiv.org/pdf/2204.09679
• 代码/Code: None

Transformers - - 2 篇

A New Dataset and Transformer for Stereoscopic Video Super-Resolution

标题：用于立体视频超分辨率的新数据集和Transformer

• 论文/Paper: http://arxiv.org/pdf/2204.10039
• 代码/Code: https://github.com/H-deep/Trans-SVSR/

Transformer Decoders with MultiModal Regularization for Cross-Modal Food Retrieval

标题：Transformer解码器，具有多模式正则化，用于跨模型食品检索

• 论文/Paper: http://arxiv.org/pdf/2204.09730
• 代码/Code: https://github.com/mshukor/TFood

多模态 / Multimodal - 1 篇

Transformer Decoders with MultiModal Regularization for Cross-Modal Food Retrieval

标题：Transformer解码器，具有多模式正则化，用于跨模型食品检索

• 论文/Paper: http://arxiv.org/pdf/2204.09730
• 代码/Code: https://github.com/mshukor/TFood

姿态估计/Pose Estimation - 1 篇

DGECN: A Depth-Guided Edge Convolutional Network for End-to-End 6D Pose Estimation

标题：DGECN：用于端到端6D姿势估计的深度引导边缘卷积网络

• 论文/Paper: http://arxiv.org/pdf/2204.09983
• 代码/Code: None

人脸相关 / Face - 1 篇

WebFace260M: A Benchmark for Million-Scale Deep Face Recognition

标题：WebFace260M：百万级深层识别的基准

• 论文/Paper: http://arxiv.org/pdf/2204.10149
• 代码/Code: None

其他/Other - 10 篇

SelfD: Self-Learning Large-Scale Driving Policies From the Web

标题：selfd：从网络中自学大规模驾驶策略

• 论文/Paper: http://arxiv.org/pdf/2204.10320
• 代码/Code: None

SmartPortraits: Depth Powered Handheld Smartphone Dataset of Human Portraits for State Estimation, Reconstruction and Synthesis

标题：SmartPortraits：深度动力手持智能手机数据集，用于估算，重建和合成

• 论文/Paper: http://arxiv.org/pdf/2204.10211
• 代码/Code: None

A case for using rotation invariant features in state of the art feature matchers

标题：在艺术状态匹配状态中使用旋转不变功能的情况

• 论文/Paper: http://arxiv.org/pdf/2204.10144
• 代码/Code: None

Toward Fast, Flexible, and Robust Low-Light Image Enhancement

标题：朝向快速，灵活，稳健的低光图像增强

• 论文/Paper: http://arxiv.org/pdf/2204.10137
• 代码/Code: https://github.com/vis-opt-group/SCI

OSSO: Obtaining Skeletal Shape from Outside

标题：OSSO：从外面获得骨骼形状

• 论文/Paper: http://arxiv.org/pdf/2204.10129
• 代码/Code: None

Is Neuron Coverage Needed to Make Person Detection More Robust?

标题：是让人们检测更强大的神经元覆盖范围吗？

• 论文/Paper: http://arxiv.org/pdf/2204.10027
• 代码/Code: None

Progressive Training of A Two-Stage Framework for Video Restoration

标题：用于视频恢复的两阶段框架的渐进训练

• 论文/Paper: http://arxiv.org/pdf/2204.09924
• 代码/Code: None

CNLL: A Semi-supervised Approach For Continual Noisy Label Learning

标题：CNLL：一种半监督持续嘈杂的标签学习方法

• 论文/Paper: http://arxiv.org/pdf/2204.09881
• 代码/Code: None

Persistent-Transient Duality in Human Behavior Modeling

标题：人类行为建模中的持续瞬态二元性

• 论文/Paper: http://arxiv.org/pdf/2204.09875
• 代码/Code: None

Self-Supervised Learning to Guide Scientifically Relevant Categorization of Martian Terrain Images

标题：自我监督学习指导科学相关分类的火星地形图像

• 论文/Paper: http://arxiv.org/pdf/2204.09854
• 代码/Code: https://github.com/TejasPanambur/mastcam