ECCV2020 收录论文汇总（持续更新中）附打包下载_hongwei yong, jianqiang huang, deyu meng, xian-she

2020极市计算机视觉开发者榜单已于2020年7月20日开赛，8月31日截止提交，基于火焰识别、电动车头盔识别、后厨老鼠识别、摔倒识别四个赛道，47000+数据集，30万奖励等你挑战！点击这里报名
极市平台（微信公众号ID：extrememart）：专注计算机视觉前沿资讯和技术干货。本文由极市平台首发，转载需获授权。

---

ECCV 2020论文已公布，本届 ECCV 共收到有效投稿5025篇，接收1361篇，其中Oral论文 104 篇，仅占 2%。

ECCV2020全部论文打包下载

ECCV2020论文开放下载后，极市第一时间进行了下载打包，由于官网下载的论文文件没有用论文名进行命名，为了方便大家学习，我们整理了所有论文名称索引：

ECCV2020全部论文下载

在极市平台公众号后台回复ECCV2020，即可获取论文打包下载链接

---

ECCV2020 Oral论文汇总

本文汇总截止今日所有Oral 论文，其中已经公布完整论文的有 47篇，按照研究方向进行了初步分类。

这些论文中有二十几篇已经公布了代码，也一并列出了。

比较有意思的是，本文列出的第一篇论文 Quaternion Equivariant Capsule Networks for 3D Point Clouds 曾被 ICLR 2020 拒稿（https://openreview.net/forum?id=B1xtd1HtPS），ECCV 2020 却给了个Oral。

下载Oral论文请点击下方链接：

https://pan.baidu.com/s/1Kd39idXZN4baUGnQrEo7hg 提取码: iyiv

后续会持续更新。

** 3D点云（ 3D Point Clouds）**

[1].Quaternion Equivariant Capsule Networks for 3D Point Clouds

作者 | Yongheng Zhao, Tolga Birdal, Jan Eric Lenssen, Emanuele Menegatti, Leonidas Guibas, Federico Tombari

单位 | Univ. Padova, TU Munich；斯坦福大学；多特蒙德工业大学等

论文 | https://arxiv.org/abs/1912.12098v2

点云补全和分类

[2].SoftpoolNet: Shape Descriptor for Point Cloud Completion and Classification

点云插值

[3].Intrinsic Point Cloud Interpolation via Dual Latent Space Navigation

作者 | Marie-Julie Rakotosaona, Maks Ovsjanikov

单位 | 巴黎综合理工学院

论文 | https://arxiv.org/abs/2004.01661

** 神经架构搜索（NAS）**

[4].MoSaNAS: Multi-Objective Surrogate-Assisted Neural Architecture Search

[5].S2DNAS: Transforming Static CNN Model for Dynamic Inference via Neural Architecture Search

作者 | Zhihang Yuan, Bingzhe Wu, Zheng Liang, Shiwan Zhao, Weichen Bi, Guangyu Sun

单位 | 北大；IBM Research

论文 | https://arxiv.org/abs/1911.07033

** 动作识别（Activity Recognition）**

[6].Empowering Relational Network by Self-Attention Augmented Conditional Random Fields for Group Activity Recognition

** 动作定位**

[7].Learning to Localize Actions from Moments

** 语义分割（Semantic Segmentation）**

[8].Synthesize then Compare: Detecting Failures and Anomalies for Semantic Segmentation

作者 | Yingda Xia, Yi Zhang, Fengze Liu, Wei Shen, Alan Yuille

单位 | 约翰斯霍普金斯大学

论文 | https://arxiv.org/abs/2003.08440

场景解析-scene parsing

[25].Semantic Flow for Fast and Accurate Scene Parsing

作者 | Xiangtai Li, Ansheng You, Zhen Zhu, Houlong Zhao, Maoke Yang, Kuiyuan Yang, Yunhai Tong

单位 | 北大；华中科技大学；DeepMotion

论文 | https://arxiv.org/abs/2002.10120

代码 | https://github.com/donnyyou/torchcv

[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-l7Gcwa3D-1594205060580)(https://bbs.cvmart.net/uploads/images/202007/08/11/gDgz6lop3D.png?imageView2/2/w/1240/h/0)]

弱监督语义分割

[26].Mining Cross-Image Semantics for Weakly Supervised Semantic Segmentation

作者 | Guolei Sun, Wenguan Wang, Jifeng Dai, Luc Van Gool

单位 | 苏黎世联邦理工学院；商汤

论文 | https://arxiv.org/abs/2007.01947

代码 | https://github.com/GuoleiSun/MCIS_wsss（尚未）

[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-GdhXiAhi-1594205060581)(https://bbs.cvmart.net/uploads/images/202007/08/11/mDuOnOfB4V.png?imageView2/2/w/1240/h/0)]

** 生成对抗网络（GAN）**

[9].House-GAN: Relational Generative Adversarial Networks for Graph-constrained House Layout Generation

作者 | Nelson Nauata, Kai-Hung Chang, Chin-Yi Cheng, Greg Mori, Yasutaka Furukawa

单位 | 西蒙弗雷泽大学；Autodesk Research

论文 | https://arxiv.org/abs/2003.06988

代码 | https://github.com/ennauata/housegan

[10].ForkGAN: Seeing into the Rainy Night

重写深度生成模型

[11].Rewriting a Deep Generative Model

[12].Aligning and Projecting Images to Class-conditional Generative Networks

无监督草图到照片的图像合成

[13].Unsupervised Sketch-to-Photo Synthesis

作者 | Runtao Liu, Qian Yu, Stella Yu

单位 | 北大；北航/伯克利；伯克利/ICSI

论文 | https://arxiv.org/abs/1909.08313

代码 | https://github.com/rt219/Unpaired-Sketch-to-Photo-Translation

[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-iBTylzY9-1594205060583)(https://bbs.cvmart.net/uploads/images/202007/08/11/kpup1J8Vkr.png?imageView2/2/w/1240/h/0)]

[14].TopoGAN: A Topology-Aware Generative Adversarial Network

** 运动捕捉**

[15].Motion Capture from Internet Videos

数据集 | https://github.com/zju3dv/iMoCap_dataset（即将）

** 人体姿态估计（ Human Pose Estimation****）**

3D人体姿态估计

[16].End-to-End Estimation of Multi-Person 3D Poses from Multiple Cameras

作者 | Hanyue Tu ，Chunyu Wang ，Wenjun Zeng

单位 | 微软亚洲研究院

论文 | https://www.microsoft.com/en-us/research/publication/end-to-end-estimation-of-multi-person-3d-poses-from-multiple-cameras/

** 目标检测（ Object Detection****）**

[17].End-to-End Object Detection with Transformers

作者 | Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov, Sergey Zagoruyko

单位 | Facebook AI

论文 | https://arxiv.org/abs/2005.12872

代码 | https://github.com/facebookresearch/detr

[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-jb5nKtHO-1594205060584)(https://bbs.cvmart.net/uploads/images/202007/08/11/tfAqZwnRkL.png?imageView2/2/w/1240/h/0)]

[18].Suppress and Balance: A Simple Gated Network for Salient Object Detection

代码 | https://github.com/Xiaoqi-Zhao-DLUT/GateNet-RGB-Saliency(即将)

[19].BorderDet: Border Feature for Dense Object Detection

** 三维重建（ 3D Reconstruction****）**

[20].Ladybird: Deep Implicit Field Based 3D Reconstruction with Sampling and Symmetry

[21].Coherent full scene 3D reconstruction from a single RGB image

3D人体重建

[22].Combining Implicit Function Learning and Parametric Models for 3D Human Reconstruction


** 多目标跟踪与分割（ MOTS****）**

[23].Segment as Points for Efficient Online Multi-Object Tracking and Segmentation

作者 | Zhenbo Xu, Wei Zhang, Xiao Tan, Wei Yang, Huan Huang, Shilei Wen, Errui Ding, Liusheng Huang

单位 | 中国科学技术大学；百度

论文 | https://arxiv.org/abs/2007.01550

代码 | https://github.com/detectRecog/PointTrack

[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-MBPMZiaS-1594205060585)(https://bbs.cvmart.net/uploads/images/202007/08/11/LWnriNBaju.png?imageView2/2/w/1240/h/0)]

** 实例分割（ Instance Segmentation****）**

[24].Conditional Convolutions for Instance Segmentation

作者 | Zhi Tian, Chunhua Shen, Hao Chen

单位 | 阿德莱德大学

论文 | https://arxiv.org/abs/2003.05664

代码 | https://github.com/aim-uofa/AdelaiDet

[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-sWEmKf9M-1594205060586)(https://bbs.cvmart.net/uploads/images/202007/08/11/YG9cImrxLn.png?imageView2/2/w/1240/h/0)]

** 视频目标分割**

[27].Learning What to Learn for Video Object Segmentation

作者 | Goutam Bhat, Felix Järemo Lawin, Martin Danelljan, Andreas Robinson, Michael Felsberg, Luc Van Gool, Radu Timofte

单位 | 苏黎世联邦理工学院；林雪平大学

论文 | https://arxiv.org/abs/2003.11540

** 数据集（ Dataset ****）**

[28].Fashionpedia: Ontology, Segmentation, and an Attribute Localization Dataset

作者 | Menglin Jia, Mengyun Shi, Mikhail Sirotenko, Yin Cui, Claire Cardie, Bharath Hariharan, Hartwig Adam, Serge Belongie

单位 | 康奈尔大学；Cornell Tech；谷歌；Hearst Magazines

论文 | https://arxiv.org/abs/2004.12276

数据集 | https://fashionpedia.github.io/home/index.html

[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-KgP78j8q-1594205060589)(https://bbs.cvmart.net/uploads/images/202007/08/11/SQJc0C8aHN.png?imageView2/2/w/1240/h/0)]

[29].Deep Fashion3D: A Dataset and Benchmark for 3D Garment Reconstruction from Single-view Images

作者 | Heming Zhu, Yu Cao, Hang Jin, Weikai Chen, Dong Du, Zhangye Wang, Shuguang Cui, Xiaoguang Han

单位 | 香港中文大学；深圳市大数据研究院；浙江大学；西安电子科技大学；腾讯-美国

论文 | https://arxiv.org/abs/2003.12753

主页 | https://kv2000.github.io/2020/03/25/deepFashion3DRevisited/

[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-pJnB41ij-1594205060589)(https://bbs.cvmart.net/uploads/images/202007/08/11/w3Z740OMnc.png?imageView2/2/w/1240/h/0)]

[30].SIZER: A Dataset and Model for Parsing 3D Clothing and Learning Size Sensitive 3D Clothing

[31].Mapillary Planet-Scale Depth Dataset

带有阅读理解功能的图像字幕数据集

[32].TextCaps: a Dataset for Image Captioning with Reading Comprehension

作者 | Oleksii Sidorov, Ronghang Hu, Marcus Rohrbach, Amanpreet Singh

单位 | FAIR；伯克利

论文 | https://arxiv.org/abs/2003.12462

代码 | https://github.com/facebookresearch/mmf/tree/project

/m4c/projects/M4C_Captioner

主页 | https://textvqa.org/textcaps
[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-wK8iUgNZ-1594205060590)(https://bbs.cvmart.net/uploads/images/202007/08/11/ThazwWgcjM.png?imageView2/2/w/1240/h/0)]

** 图像字幕（ Image Captioning **）

[33].Compare and Reweight: Distinctive Image Captioning Using Similar Images Sets

** 人体运动预测（ Human Motion Prediction ****）**

[34].Long-term Human Motion Prediction with Scene Context

作者 | Zhe Cao , Hang Gao , Karttikeya Mangalam , Qi-Zhi Cai , Minh Vo , and Jitendra Malik

单位 | 伯克利；南京大学；Facebook Reality Lab

论文 | https://people.eecs.berkeley.edu/~zhecao/hmp/preprint.pdf

主页 | https://people.eecs.berkeley.edu/~zhecao/hmp/index.html

[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-oFwszsgt-1594205060591)(https://bbs.cvmart.net/uploads/images/202007/08/11/q8pfPd1r3F.png?imageView2/2/w/1240/h/0)]

** 场景识别**

[35].ReferIt3D: Neural Listeners for Fine-Grained 3D Object Identification in Real-World Scenes

** 视图合成**

[36].NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis

作者 | Ben Mildenhall, Pratul P. Srinivasan, Matthew Tancik, Jonathan T. Barron, Ravi Ramamoorthi, Ren Ng

单位 | 伯克利；谷歌；加利福尼亚大学圣迭戈分校

论文 | https://arxiv.org/abs/2003.08934

代码 | https://github.com/bmild/nerf

主页 | https://www.matthewtancik.com/nerf
[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-HLlubLOL-1594205060592)(https://bbs.cvmart.net/uploads/images/202007/08/11/l28xOrEaKY.png?imageView2/2/w/1240/h/0)]

6DoF视频视图合成

[37].MatryODShka: Real-time 6DoF Video View Synthesis using Multi-Sphere Images

** 深度估计（Depth Estimation）**

[38].Du^2Net: Learning Depth Estimation from Dual-Cameras and Dual-Pixels

作者 | Yinda Zhang, Neal Wadhwa, Sergio Orts-Escolano, Christian Häne, Sean Fanello, Rahul Garg

单位 | 谷歌

论文 | https://arxiv.org/abs/2003.14299
[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-9JNK7Kc1-1594205060592)(https://bbs.cvmart.net/uploads/images/202007/08/11/WqF4nSuFU4.png?imageView2/2/w/1240/h/0)]

** 少样本学习（Few-Shot learning）**

[39].Model-Agnostic Boundary-Adversarial Sampling for Test-Time Generalization in Few-Shot learning

[40].Prototype Rectification for Few-Shot Learning

作者 | Jinlu Liu, Liang Song, Yongqiang Qin

单位 | 创新奇智

论文 | https://arxiv.org/abs/1911.10713

[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-XYoZDCiZ-1594205060594)(https://bbs.cvmart.net/uploads/images/202007/08/11/DQF8fijCPj.png?imageView2/2/w/1240/h/0)]


** 无监督学习（Unsupervised learning）**

[41].Content-Aware Unsupervised Deep Homography Estimation

作者 | Jirong Zhang, Chuan Wang, Shuaicheng Liu, Lanpeng Jia, Jue Wang, Ji Zhou

单位 | 电子科技大学

论文 | https://arxiv.org/abs/1909.05983

代码 | https://github.com/JirongZhang/DeepHomography

[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-yXFqsR1z-1594205060595)(https://bbs.cvmart.net/uploads/images/202007/08/11/OTadBZzWgT.png?imageView2/2/w/1240/h/0)]
[42].Visual Memorability for Robotic Interestingness Prediction via Unsupervised Online Learning

作者 | Chen Wang, Wenshan Wang, Yuheng Qiu, Yafei Hu, Sebastian Scherer

单位 | 卡内基梅隆大学

论文 | https://arxiv.org/abs/2005.08829v1

视频 | https://www.youtube.com/watch?v=Gy15Hx6qiGE

在线视频：
https://v.qq.com/x/page/f3111e7f5ci.html?start=1

** 监督学习/自监督/半监督**

[43].Learning Feature Descriptors using Camera Pose Supervision

作者 | Qianqian Wang, Xiaowei Zhou, Bharath Hariharan, Noah Snavely

单位 | 康奈尔大学；Cornell Tech；浙江大学

论文 | https://arxiv.org/abs/2004.13324

主页 | https://qianqianwang68.github.io/DescfromPose/

实时互动演示视频：
https://v.qq.com/x/page/o3111tx6o1h.html

方法：
[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-ToDDd0I3-1594205060596)(https://bbs.cvmart.net/uploads/images/202007/08/11/ZJhZq9W6cj.png?imageView2/2/w/1240/h/0)]

结果：

[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-ZxxHfyoS-1594205060597)(https://bbs.cvmart.net/uploads/images/202007/08/11/8Q4nVOwk15.png?imageView2/2/w/1240/h/0)]

[44].Appearance Consensus Driven Self-Supervised Human Mesh Recovery

[45].TCGM: An Information-Theoretic Framework for Semi-Supervised Multi-Modality Learning

** 人物交互（Human-Object Interaction）**

[46].Forecasting Human-Object Interaction: Joint Prediction of Motor Attention and Actions in First Person Video

作者 | Miao Liu, Siyu Tang, Yin Li, James Rehg

单位 | 佐治亚理工学院；威斯康星大学麦迪逊分校等

论文 | https://arxiv.org/abs/1911.10967
[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-VfDfGqG6-1594205060599)(https://bbs.cvmart.net/uploads/images/202007/08/11/1iz1XWPyVT.png?imageView2/2/w/1240/h/0)]

** 人员重识别（Person Re-Identification ）**

[47].Joint Disentangling and Adaptation for Cross-Domain Person Re-Identification

[48].Appearance-Preserving 3D Convolution for Video-based Person Re-identification

** 车辆重识别（Vehicle Re-Identification ）**

[49].Orientation-aware Vehicle Re-identification with Semantics-guided Part Attention Network

** 视觉问答（VQA ）**

[50].A Competence-aware Curriculum for Visual Concepts Learning via Question Answering

作者 | Qing Li, Siyuan Huang, Yining Hong, Song-Chun Zhu

单位 | 加州大学洛杉矶分校

论文 | https://arxiv.org/abs/2007.01499
[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-bwooWVKB-1594205060600)(https://bbs.cvmart.net/uploads/images/202007/08/11/p6ILuVLG39.png?imageView2/2/w/1240/h/0)]

** 视频压缩（Video Compression ）**

[51].Improving Deep Video Compression by Resolution-adaptive Flow Coding

[52].Content Adaptive and Error Propagation Aware Deep Video Compression

作者 | Guo Lu, Chunlei Cai, Xiaoyun Zhang, Li Chen, Wanli Ouyang, Dong Xu, Zhiyong Gao

单位 | 上海交通大学；悉尼大学；悉尼大学/商汤CV研究小组

论文 | https://arxiv.org/abs/2003.11282
[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-iGGFDnmK-1594205060601)(https://bbs.cvmart.net/uploads/images/202007/08/11/KTChgaNmWX.png?imageView2/2/w/1240/h/0)]

** 图像恢复（Image Restoration）**

[53].Exploiting Deep Generative Prior for Versatile Image Restoration and Manipulation

作者 | Xingang Pan, Xiaohang Zhan, Bo Dai, Dahua Lin, Chen Change Loy, Ping Luo

单位 | 香港中文大学；南洋理工大学 ；香港大学

论文 | https://arxiv.org/abs/2003.13659

代码 | https://github.com/XingangPan/deep-generative-prior

[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-k7tsWj40-1594205060605)(https://bbs.cvmart.net/uploads/images/202007/08/11/QaYxsSoX5V.png?imageView2/2/w/1240/h/0)]

** 图像修复（inpainting）**

[54].Rethinking image inpainting via a mutual encoder-decoder with feature equalization

代码 | https://github.com/KumapowerLIU/ECCV2020oralRethinking-Image-Inpainting-via-a-Mutual-Encoder-Decoder-with-Feature-Equalizations（尚未）

** 图像检索（Image Retrieval）**

[55].ExchNet: A Unified Hashing Network for Large-Scale Fine-Grained Image Retrieval

对抗学习 + 图像检索

[56].Targeted Attack for Deep Hashing based Retrieval

作者 | Jiawang Bai, Bin Chen, Yiming Li, Dongxian Wu, Weiwei Guo, Shu-tao Xia, En-hui Yang

单位 | 清华大学；鹏城实验室；Vivo；滑铁卢大学

论文 | https://arxiv.org/abs/2004.07955

[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-emjM4RaQ-1594205060605)(https://bbs.cvmart.net/uploads/images/202007/08/11/vSMGEtIxaO.png?imageView2/2/w/1240/h/0)]

图像检索，深度局部特征聚合

[57].Learning and aggregating deep local descriptors for instance-level recognition

** 遥感与航空影像处理与识别**

[58].Synthesis and Completion of Facades from Satellite Imagery

** 计算成像**

成像技术

[59].Diffraction Line Imaging

光场重建

[60].Deep Spatial-angular Regularization for Compressive Light Field Reconstruction over Coded Apertures

** 光流估计**

[61].RAFT: Recurrent All-Pairs Field Transforms for Optical Flow

作者 | Zachary Teed, Jia Deng

单位 | 普林斯顿大学

论文 | https://arxiv.org/abs/2003.12039

代码 | https://github.com/princeton-vl/RAFT

[62].What Matters in Unsupervised Optical Flow

作者 | Rico Jonschkowski, Austin Stone, Jonathan T. Barron, Ariel Gordon, Kurt Konolige, Anelia Angelova

单位 | 谷歌

论文 | https://arxiv.org/abs/2006.04902

代码 | https://github.com/google-research/google-research/tree/master/uflow

** PnP问题求解**

[63].Solving the Blind Perspective-n-Point Problem End-To-End With Robust Differentiable Geometric Optimization

[64].A Consistently Fast and Globally Optimal Solution to the Perspective-n-Point Problem

** 网络剪枝、量化、加速**

深度神经网络的训练后分段线性量化

[65].Post-Training Piecewise Linear Quantization for Deep Neural Networks

作者 | Jun Fang, Ali Shafiee, Hamzah Abdel-Aziz, David Thorsley, Georgios Georgiadis, Joseph Hassoun

单位 | 三星；微软

论文 | https://arxiv.org/abs/2002.00104

网络剪枝

[66].EagleEye: Fast Sub-net Evaluation for Efficient Neural Network Pruning

作者 | Bailin Li, Bowen Wu, Jiang Su, Guangrun Wang, Liang Lin

单位 | Dark Matter AI Inc；中山大学

论文 | https://arxiv.org/abs/2007.02491

代码 | https://github.com/anonymous47823493/EagleEye
[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-stl64kZN-1594205060609)(https://bbs.cvmart.net/uploads/images/202007/08/11/Lq1ljc25NC.png?imageView2/2/w/1240/h/0)]

网络推断加速，使用采样-插值方法减少CNN空间信息冗余

[67].Spatially Adaptive Inference with Stochastic Feature Sampling and Interpolation

作者 | Zhenda Xie, Zheng Zhang, Xizhou Zhu, Gao Huang, Stephen Lin

单位 | 清华大学；微软亚洲研究院；中国科学技术大学

论文 | https://arxiv.org/abs/2003.08866
[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-T7jQp2ap-1594205060609)(https://bbs.cvmart.net/uploads/images/202007/08/11/f5dcN6g8pS.png?imageView2/2/w/1240/h/0)]

** 网络优化、正则化**

优化技术

[68].Gradient Centralization: A New Optimization Technique for Deep Neural Networks

作者 | Hongwei Yong, Jianqiang Huang, Xiansheng Hua, Lei Zhang

单位 | 香港理工大学；阿里达摩院

论文 | https://arxiv.org/abs/2004.01461

代码 | https://github.com/Yonghongwei/Gradient-Centralization

深度神经网络的逐层条件分析

[69].Layer-wise Conditioning Analysis in Exploring the Learning Dynamics of DNNs

作者 | Lei Huang, Jie Qin, Li Liu, Fan Zhu, Ling Shao

单位 | IIAI

论文 | https://arxiv.org/abs/2002.10801

代码 | https://github.com/huangleiBuaa/LayerwiseCA
[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-1eaVVMFd-1594205060611)(https://bbs.cvmart.net/uploads/images/202007/08/11/6KHHH1BZqi.png?imageView2/2/w/1240/h/0)]

[70].Regularization with Latent Space Virtual Adversarial Training

** Structure-from-Motion**

[71].DeepSFM: Structure From Motion Via Deep Bundle Adjustment

作者 | Xingkui Wei, Yinda Zhang, Zhuwen Li, Yanwei Fu, Xiangyang Xue

单位 | 复旦大学；谷歌；Nuro, Inc

论文 | https://arxiv.org/abs/1912.09697

[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-zXpn6faN-1594205060613)(https://bbs.cvmart.net/uploads/images/202007/08/11/U9hwmlz3uU.png?imageView2/2/w/1240/h/0)]

[72].Privacy Preserving Structure-from-Motion

** 局部特征提取**

局部特征提精

[73].Multi-View Optimization of Local Feature Geometry

作者 | Mihai Dusmanu, Johannes L. Schönberger, Marc Pollefeys

单位 | 苏黎世联邦理工学院；微软

论文 | https://arxiv.org/abs/2003.08348

代码 | https://github.com/mihaidusmanu/local-feature-refinement
[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-0e1fw0bI-1594205060613)(https://bbs.cvmart.net/uploads/images/202007/08/11/uJAFFDfw1J.png?imageView2/2/w/1240/h/0)]

局部特征描述

[74].Online Invariance Selection for Local Feature Descriptors

** 人脸**

基于模型的密集人脸配准技术

[75].“Look Ma, no landmarks!” - Unsupervised, model-based dense face alignment

人脸老化

[76].Hierarchical Face Aging through Disentangled Latent Characteristics

** 自动驾驶**

人类活动预测

[77].Adversarial Generative Grammars for Human Activity Prediction

运动预测

[78].Learning Lane Graph Representations for Motion Forecasting

自动驾驶

车辆通信的联合感知与预测

[79].V2VNet: Vehicle-to-Vehicle Communication for Joint Perception and Prediction

作者主页 | https://zswang666.github.io/#（未公布）

轨迹预测

[80].It is not the Journey but the Destination: Endpoint Conditioned Trajectory Prediction

作者 | Karttikeya Mangalam, Harshayu Girase, Shreyas Agarwal, Kuan-Hui Lee, Ehsan Adeli, Jitendra Malik, Adrien Gaidon

单位 | 伯克利；丰田；斯坦福大学

论文 | https://arxiv.org/abs/2004.02025

主页 | https://karttikeya.github.io/publication/htf/

代码 | 即将

[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-Ycuu3dg8-1594205060614)(https://bbs.cvmart.net/uploads/images/202007/08/11/4WBfMha6o9.png?imageView2/2/w/1240/h/0)]

** 立体匹配**

[81].Domain-invariant Stereo Matching Networks

作者 | Feihu Zhang, Xiaojuan Qi, Ruigang Yang, Victor Prisacariu, Benjamin Wah, Philip Torr

单位 | 牛津大学；百度；香港中文大学

论文 | https://arxiv.org/abs/1911.13287

代码 | https://github.com/feihuzhang/DSMNet
[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-kfyaqk0a-1594205060615)(https://bbs.cvmart.net/uploads/images/202007/08/11/IFpYMu26wA.png?imageView2/2/w/1240/h/0)]

** 单RGB图像的手部网格建模**

[82].DeepHandMesh: Weakly-supervised Deep Encoder-Decoder Framework for High-fidelity Hand Mesh Modeling from a Single RGB Image

** 流式图像理解**

[83].Towards Streaming Image Understanding

作者 | Mengtian Li, Yu-Xiong Wang, Deva Ramanan

单位 | 卡内基梅隆；Argo AI

论文 | https://arxiv.org/abs/2005.10420

主页 | https://www.cs.cmu.edu/~mengtial/proj/streaming/

https://v.qq.com/x/page/j31110iijxz.html

[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-taHI5Bmy-1594205060615)(https://bbs.cvmart.net/uploads/images/202007/08/11/lhzFJXPrAY.png?imageView2/2/w/1240/h/0)]

** 度量保留的可变形3D形状表示**

[84].LIMP: Learning Latent Shape Representations with Metric Preservation Priors

作者 | Luca Cosmo, Antonio Norelli, Oshri Halimi, Ron Kimmel, Emanuele Rodolà

单位 | 罗马大学；University of Lugano；以色列理工学院

论文 | https://arxiv.org/abs/2003.12283

** 开放集识别**

[85].Hybrid Models for Open Set Recognition

作者 | Hongjie Zhang, Ang Li, Jie Guo, Yanwen Guo

单位 | 南京大学；DeepMind

论文 | https://arxiv.org/abs/2003.12506

** 3D表面拟合**

[86].DeepFit: 3D Surface Fitting by Neural Network Weighted Least Squares

作者 | Yizhak Ben-Shabat, Stephen Gould

单位 | 澳大利亚国立大学

论文 | https://arxiv.org/abs/2003.10826

** 6D物体姿态估计（6D Object Pose Estimation）**

[87].Self6D: Self-Supervised Monocular 6D Object Pose Estimation

作者 | Gu Wang, Fabian Manhardt, Jianzhun Shao, Xiangyang Ji, Nassir Navab, Federico Tombari

单位 | 清华大学；慕尼黑工业大学；谷歌

论文 | https://arxiv.org/abs/2004.06468
[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-OO1KVd9h-1594205060620)(https://bbs.cvmart.net/uploads/images/202007/08/11/Urtjh2t9pg.png?imageView2/2/w/1240/h/0)]

** 图像缩放 （Image Rescaling）**

[88].Invertible Image Rescaling

作者 | Mingqing Xiao, Shuxin Zheng, Chang Liu, Yaolong Wang, Di He, Guolin Ke, Jiang Bian, Zhouchen Lin, Tie-Yan Liu

单位 | 北大；微软亚洲研究院；多伦多大学

论文 | https://arxiv.org/abs/2005.05650

代码 | https://github.com/pkuxmq/Invertible-Image-Rescaling（即将）

解读 | https://zhuanlan.zhihu.com/p/150340687
[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-XzQCErOO-1594205060622)(https://bbs.cvmart.net/uploads/images/202007/08/11/4JVzUxpVP9.png?imageView2/2/w/1240/h/0)]
** 远程生理信号监控**

远程生理测量：通过交叉验证特征解耦的方法

[89].Video-based Remote Physiological Measurement via Cross-verified Feature Disentangling

** 网络结构设计**

自适应网络宽度和分辨率学习

[90].MutualNet: Adaptive ConvNet via Mutual Learning from Network Width and Resolution

作者 | Taojiannan Yang, Sijie Zhu, Chen Chen, Shen Yan, Mi Zhang, Andrew Willis

单位 | 北卡罗来纳大学夏洛特分校；密歇根州立大学

论文 | https://arxiv.org/abs/1909.12978

代码 | https://github.com/taoyang1122/MutualNet
[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-WKVZDvRc-1594205060623)(https://bbs.cvmart.net/uploads/images/202007/08/11/0vjYl2kRAg.png?imageView2/2/w/1240/h/0)]

** 其它**

[91].In-Home Daily-Life Captioning Using Radio Signals

[92].Self-Challenging Improves Cross-Domain Generalization

作者 | Zeyi Huang, Haohan Wang, Eric P. Xing, Dong Huang

单位 | 卡内基梅隆大学

论文 | https://arxiv.org/abs/2007.02454
[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-Ts4vCBOT-1594205060624)(https://bbs.cvmart.net/uploads/images/202007/08/11/hMPjuMkAk1.png?imageView2/2/w/1240/h/0)]

多任务学习增强对抗鲁棒性

[93].Multi-task Learning Increases Adversarial Robustness

[94].Efficient Model Fitting by Combining Lifted Optimization with Phong Surface Models

从单图像学习立体信息

[95].Learning Stereo from Single Images

[96].Towards Automated Testing and Robustification by Semantic Adversarial Data Generation

持续学习-Continual Learning

[97].Greedy Sampler and Dumb Learner: A Surprisingly Effective Approach for Continual Learning

可解释CNN

[98].Training Interpretable Convolutional Neural Networks by Differentiating Class-specific Filters

跨域级联深度翻译

[99].Cross-Domain Cascaded Deep Translation

使神经网络能够抵抗各种图像损坏的简单方法

[100].A simple way to make neural networks robust against diverse image corruptions

使用自然语言描述纹理

[101].Describing Textures using Natural Language

具有推理能力的注意力模型

[102].AiR: Attention with Reasoning Capability

代码 | https://github.com/szzexpoi/AiR

[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-x1jQfOdY-1594205060625)(https://bbs.cvmart.net/uploads/images/202007/08/11/aqrrzj7Cj5.png?imageView2/2/w/1240/h/0)]

[103].Crowdsampling the Plenoptic Function

[104].恢复视频监控的可见光颜色

Learn to Recover Visible Color for Video Surveillance in a Day

文章来源： OpenCV中文网@微信公众号