计算机视觉/图像处理学术速递[01.08]

格林先生MrGreen arXiv每日学术速递
www.arxivdaily.com上线啦，论文摘要、多学科、收藏、评论、搜索……，点击文末“阅读原文”即可访问
cs.CV 方向，今日共计41篇

[检测分类相关]：

【1】 Self-Attention Based Context-Aware 3D Object Detection

标题：基于自我注意的上下文感知三维目标检测

作者：Prarthana Bhattacharyya,Chengjie Huang,Krzysztof Czarnecki
机构：University of Waterloo, Canada, RGB Image, Without Self-attention, With Self-attention (ours)
备注：17 pages, 9 figures
链接：arxiv.org/abs/2101.02672

【2】 MSED: a multi-modal sleep event detection model for clinical sleep analysis

标题：MSED：一种用于临床睡眠分析的多模式睡眠事件检测模型

作者：Alexander Neergaard Olesen,Poul Jennum,Emmanuel Mignot,Helge B. D. Sorensen
机构：January
备注：20 pages, 6 figures
链接：arxiv.org/abs/2101.02530

【3】 Active learning for object detection in high-resolution satellite images

标题：基于主动学习的高分辨率卫星图像目标检测

作者：Alex Goupilleau,Tugdual Ceillier,Marie-Caroline Corbineau
机构：Preligens (ex-Earthcube), Paris, France
备注：None
链接：arxiv.org/abs/2101.02480

【4】 Practical Evaluation of Out-of-Distribution Detection Methods for Image Classification

标题：图像分类中非分布检测方法的实用化评价

作者：Engkarat Techapanurak,Takayuki Okatani
机构：Tohoku University, FRIKEN Center for AIP
链接：arxiv.org/abs/2101.02447

【5】 Progressive Self-Guided Loss for Salient Object Detection

标题：用于显着目标检测的渐进式自导损失算法

作者：Sheng Yang,Weisi Lin,Guosheng Lin,Qiuping Jiang,Zichuan Liu
备注：In submission
链接：arxiv.org/abs/2101.02412

【6】 Low-cost and high-performance data augmentation for deep-learning-based skin lesion classification

标题：基于深度学习的低成本高性能皮肤病变分类数据增强

作者：Shuwei Shen,Mengjuan Xu,Fan Zhang,Pengfei Shao,Honghong Liu,Liang Xu,Chi Zhang,Peng Liu,Zhihong Zhang,Peng Yao,Ronald X. Xu
机构：First Affiliated Hospital, University of Science and Technology of China, Hefei , China, The Ohio State University, Columbus, OH , USA
备注：8 pages, 5 figures
链接：arxiv.org/abs/2101.02353

【7】 LightLayers: Parameter Efficient Dense and Convolutional Layers for Image Classification

标题：LightLayers：用于图像分类的参数高效密集卷积层

作者：Debesh Jha,Anis Yazidi,Michael A. Riegler,Dag Johansen,Håvard D. Johansen,Pål Halvorsen
机构： SimulaMet, Norway, UIT The Arctic University of Norway, Oslo Metropolitan University, Norway
链接：arxiv.org/abs/2101.02268

【8】 Object Detection for Understanding Assembly Instruction Using Context-aware Data Augmentation and Cascade Mask R-CNN

标题：基于上下文感知数据增强和级联掩码R-CNN的汇编指令理解目标检测

作者：J. Lee,S. Lee,S. Back,S. Shin,K. Lee
链接：arxiv.org/abs/2101.02509

【9】 OAAE: Adversarial Autoencoders for Novelty Detection in Multi-modal Normality Case via Orthogonalized Latent Space

标题：OAAE：基于正交化潜在空间的多模态正态新颖性检测对抗性自动编码器

作者：Sungkwon An,Jeonghoon Kim,Myungjoo Kang,Shahbaz Razaei,Xin Liu
机构： Computational Science and Technology, Seoul National University, Korea, Seoul National University,Korea, University of California, Davis, CA
备注：Accepted to AAAI 2021 Workshop: Towards Robust, Secure and Efficient Machine Learning
链接：arxiv.org/abs/2101.02358

【10】 Single Shot Multitask Pedestrian Detection and Behavior Prediction

标题：单镜头多任务行人检测与行为预测

作者：Prateek Agrawal,Pratik Prabhanjan Brahma
机构：Volkswagen Group, Innovation Center California, Belmont, California
备注：6 pages, 3 figures, Neurips 2020 ML4AD workshop
链接：arxiv.org/abs/2101.02232

[分割/语义相关]：

【1】 Boundary-Aware Geometric Encoding for Semantic Segmentation of Point Clouds

标题：边界感知几何编码在点云语义分割中的应用

作者：Jingyu Gong,Jiachen Xu,Xin Tan,Jie Zhou,Yanyun Qu,Yuan Xie,Lizhuang Ma
机构：Shanghai Jiao Tong University, Shanghai, China, East China Normal University, Shanghai, China, City University of Hong Kong, HKSAR, China, Xiamen University, Fujian, China
备注：Accepted by AAAI2021
链接：arxiv.org/abs/2101.02381

【2】 Diminishing Uncertainty within the Training Pool: Active Learning for Medical Image Segmentation

标题：减少训练池中的不确定性：医学图像分割的主动学习

作者：Vishwesh Nath,Dong Yang,Bennett A. Landman,Daguang Xu,Holger R. Roth
机构：NVIDIA, Bethesda, USA
备注：19 pages, 13 figures, Transactions of Medical Imaging
链接：arxiv.org/abs/2101.02323

【3】 Dual-Teacher++: Exploiting Intra-domain and Inter-domain Knowledge with Reliable Transfer for Cardiac Segmentation

标题：双师++：利用可靠传输的域内和域间知识进行心脏分割

作者：Kang Li,Shujun Wang,Lequan Yu,Pheng-Ann Heng
备注：Accepted by TMI
链接：arxiv.org/abs/2101.02375

[人脸相关]：

【1】 A Large-Scale, Time-Synchronized Visible and Thermal Face Dataset

标题：一种大规模、时间同步的可见光和热像面数据集

作者：Domenick Poster,Matthew Thielke,Robert Nguyen,Srinivasan Rajaraman,Xing Di,Cedric Nimpa Fondje,Vishal M. Patel,Nathaniel J. Short,Benjamin S. Riggan,Nasser M. Nasrabadi,Shuowen Hu
机构： West Virginia University, Evansdale Dr., Morgantown, WV , DEVCOM Army Research Laboratory, Powder Mill Rd., Adelphi, MD , Booz Allen Hamilton, Grennsboro Dr., McLean, VA , Johns Hopkins University, N. Charles Street, Baltimore, MD , University of Nebraska-Lincoln, R St, Lincoln, NE
链接：arxiv.org/abs/2101.02637

[GAN/对抗式/生成式相关]：

【1】 GAN-Control: Explicitly Controllable GANs

标题：GaN-Control：显式可控GAN

作者：Alon Shoshan,Nadav Bhonker,Igor Kviatkovsky,Gerard Medioni
机构：Amazon One
链接：arxiv.org/abs/2101.02477

【2】 VOGUE: Try-On by StyleGAN Interpolation Optimization

标题：Vogue：StyleGan插值优化试穿

作者：Kathleen M Lewis,Srivatsan Varadharajan,Ira Kemelmacher-Shlizerman
机构：Google Research ,MIT CSAIL ,University of Washington, Person, Garment, Shirt try-on, Pants try-on
链接：arxiv.org/abs/2101.02285

【3】 Robust Text CAPTCHAs Using Adversarial Examples

标题：使用对抗性示例的健壮文本验证码

作者：Rulin Shao,Zhouxing Shi,Jinfeng Yi,Pin-Yu Chen,Cho-Jui Hsieh
机构：Xi'an Jiaotong University University of California, Los Angeles JD AI Research, IBM Research
链接：arxiv.org/abs/2101.02483

【4】 VHS to HDTV Video Translation using Multi-task Adversarial Learning

标题：基于多任务对抗性学习的VHS到HDTV视频翻译

作者：Hongming Luo,Guangsen Liao,Xianxu Hou,Bozhi Liu,Fei Zhou,Guoping Qiu
机构：College of Electronics and Information Engineering Shenzhen University, Shenzhen,China, Guangdong Key Laboratory of Intelligent Information Processing, Shenzhen, China, Guangdong Laboratory of Artificial Intelligence and Digital Economy (SZ), Shenzhen Institute of Artificial Intelligence and Robotics for Society, Shenzhen, University of Nottingham, Nottingham, UK
备注：MMM2020 final version
链接：arxiv.org/abs/2101.02384

[行为/时空/光流/姿态/运动]：

【1】 PandaNet : Anchor-Based Single-Shot Multi-Person 3D Pose Estimation

标题：PandaNet：基于锚点的单炮多人三维位姿估计

作者：Abdallah Benzine,Florian Chabot,Bertrand Luvison,Quoc Cong Pham,Cahterine Achrd
机构：CEA LIST Vision and Learning Lab for Scene Analysis, Sorbonne University, CNRS, Institute for Intelligent Systems and Robotics
链接：arxiv.org/abs/2101.02471

【2】 Associated Spatio-Temporal Capsule Network for Gait Recognition

标题：用于步态识别的关联时空胶囊网络

作者：Aite Zhao,Junyu Dong,Jianbo Li,Lin Qi,Huiyu Zhou
链接：arxiv.org/abs/2101.02458

【3】 Safety-Oriented Pedestrian Motion and Scene Occupancy Forecasting

标题：面向安全的行人运动与场景占有率预测

作者：Katie Luo,Sergio Casas,Renjie Liao,Xinchen Yan,Yuwen Xiong,Wenyuan Zeng,Raquel Urtasun
机构：Uber ATGI, University of Toronto, Cornell University
链接：arxiv.org/abs/2101.02385

[半/弱/无监督相关]：

【1】 Self-Supervised Pretraining of 3D Features on any Point-Cloud

标题：任意点云上三维要素的自监督预训练

作者：Zaiwei Zhang,Rohit Girdhar,Armand Joulin,Ishan Misra
机构： Facebook AI Research ,The University of Texas at Austin
链接：arxiv.org/abs/2101.02691

[跟踪相关]：

【1】 TrackFormer: Multi-Object Tracking with Transformers

标题：TrackFormer：使用Transformers进行多目标跟踪

作者：Tim Meinhardt,Alexander Kirillov,Laura Leal-Taixe,Christoph Feichtenhofer
机构：ITechnical University of Munich, Facebook AI Research(FAIR)
备注：Tech. report
链接：arxiv.org/abs/2101.02702

[迁移学习/domain/主动学习/自适应]：

【1】 Partial Domain Adaptation Using Selective Representation Learning For Class-Weight Computation

标题：基于选择表示学习的部分区域自适应类权计算

作者：Sandipan Choudhuri,Riti Paul,Arunabha Sen,Baoxin Li,Hemanth Venkateswara
机构：CIDSE, Arizona State University
链接：arxiv.org/abs/2101.02275

[Re-id相关]：

【1】 HAVANA: Hierarchical and Variation-Normalized Autoencoder for Person Re-identification

标题：哈瓦那：层次化和变异归一化自动编码器用于人的重新识别

作者：Jiawei Ren,Xiao Ma,Chen Xu,Haiyu Zhao,Shuai Yi
备注：Manuscript
链接：arxiv.org/abs/2101.02568

[点云]：

【1】 Efficient 3D Point Cloud Feature Learning for Large-Scale Place Recognition

标题：面向大规模位置识别的高效三维点云特征学习

作者：Le Hui,Mingmei Cheng,Jin Xie,Jian Yang
备注：Project page: this https URL
链接：arxiv.org/abs/2101.02374

[3D/3D重建等相关]：

【1】 Where2Act: From Pixels to Actions for Articulated 3D Objects

标题：Where 2Act：从像素到关节3D对象的动作

作者：Kaichun Mo,Leonidas Guibas,Mustafa Mukadam,Abhinav Gupta,Shubham Tulsiani
机构：Stanford University ,Facebook AI Research
链接：arxiv.org/abs/2101.02692

[其他视频相关]：

【1】 Learning Temporal Dynamics from Cycles in Narrated Video

标题：从旁白视频中的循环学习时间动力学

作者：Dave Epstein,Jiajun Wu,Cordelia Schmid,Chen Sun
机构：Cordelia schmid, UC Berkeley Stanford University, Google, Brown University
链接：arxiv.org/abs/2101.02337

[其他]：

【1】 PVA: Pixel-aligned Volumetric Avatars

标题：PVA：像素对齐的体积化身

作者：Amit Raj,Michael Zollhoefer,Tomas Simon,Jason Saragih,Shunsuke Saito,James Hays,Stephen Lombardi
机构： Georgia Institute of Technology , Facebook Reality Labs
备注：Project page located at this https URL
链接：arxiv.org/abs/2101.02697

【2】 L2PF -- Learning to Prune Faster

标题：L2PF--学会更快地修剪

作者：Manoj-Rohit Vemparala,Nael Fasfous,Alexander Frickenstein,Mhd Ali Moraly,Aquib Jamal,Lukas Frickenstein,Christian Unger,Naveen-Shankar Nagaraja,Walter Stechele
机构：indicates equal contributions, BMW Autonomous Driving, Technical University of Munich
链接：arxiv.org/abs/2101.02663

【3】 More Reliable AI Solution: Breast Ultrasound Diagnosis Using Multi-AI Combination

标题：更可靠的人工智能解决方案：使用多人工智能组合的乳房超声诊断

作者：Jian Dai,Shuge Lei,Licong Dong,Xiaona Lin,Huabin Zhang,Desheng Sun,Kehong Yuan
机构：Yuanl, Graduate School at Shenzhen, Tsinghua University, Shenzhen, China, Computer Science and Engineering, University of South Carolina, SC, United States, Shenzhen Hospitalof Peking University, Shenzhen, China, Beijing Tsinghua Changgeng Hospital, Beijing, China
备注：12 pages, 6 figures, 6 tables
链接：arxiv.org/abs/2101.02639

【4】 Learning Anthropometry from Rendered Humans

标题：从渲染的人类中学习人体测量学

作者：Song Yan,Joni-Kristian Kämäräinen
机构：Computer Sciences, Tampere University, Finland
链接：arxiv.org/abs/2101.02515

【5】 Bridging In- and Out-of-distribution Samples for Their Better Discriminability

标题：桥接分布内和分布外样本以获得更好的区分性

作者：Engkarat Techapanurak,Anh-Chuong Dang,Takayuki Okatani
机构：Tohoku University ,RIKEN Center for AIP, Sendai,-, Japan
链接：arxiv.org/abs/2101.02500

【6】 The joint role of geometry and illumination on material recognition

标题：几何和光照在材料识别中的联合作用

作者：Manuel Lagunas,Ana Serrano,Diego Gutierrez,Belen Masia
机构：Universidad de Zaragoza, I,A, Max Planck Institute for, Zaragoza, Spain, Informatics
备注：15 pages, 16 figures, Accepted to the Journal of Vision
链接：arxiv.org/abs/2101.02496

【7】 Deep Learning Methods for Vessel Trajectory Prediction based on Recurrent Neural Networks

标题：基于递归神经网络的船舶航迹预测深度学习方法

作者：Samuele Capobianco,Leonardo M. Millefiori,Nicola Forti,Paolo Braca,Peter Willett
备注：Submitted to Transactions on Aerospace and Electronic Systems, 14 pages, 8 figures
链接：arxiv.org/abs/2101.02486

【8】 Multimodal Gait Recognition for Neurodegenerative Diseases

标题：神经退行性疾病的多模态步态识别

作者：Aite Zhao,Jianbo Li,Junyu Dong,Lin Qi,Qianni Zhang,Ning Li,Xin Wang,Huiyu Zhou
链接：arxiv.org/abs/2101.02469

【9】 Multi-scale Information Assembly for Image Matting

标题：用于图像遮片的多尺度信息拼接

作者：Yu Qiao,Yuhao Liu,Qiang Zhu,Xin Yang,Yuxin Wang,Qiang Zhang,Xiaopeng Wei
机构：College of Computer Science, Dalian University of Technology
备注：10 pages, 6 figures
链接：arxiv.org/abs/2101.02391

【10】 Who's a Good Boy? Reinforcing Canine Behavior using Machine Learning in Real-Time

标题：谁是好孩子？利用机器学习实时增强犬的行为

作者：Jason Stock,Tom Cavey
机构：Dept. Computer Science, Colorado State University
备注：8 pages, 6 figures
链接：arxiv.org/abs/2101.02380

【11】 Distribution-Free, Risk-Controlling Prediction Sets

标题：无分布、可控制风险的预测集

作者：Stephen Bates,Anastasios Angelopoulos,Lihua Lei,Jitendra Malik,Michael I. Jordan
机构：January
备注：Project website available at this http URL and codebase available at this https URL
链接：arxiv.org/abs/2101.02703

【12】 From Learning to Relearning: A Framework for Diminishing Bias in Social Robot Navigation

标题：从学习到再学习：减少社会机器人导航偏差的框架

作者：Juana Valeria Hurtado,Laura Londoño,Abhinav Valada
链接：arxiv.org/abs/2101.02647

【13】 Few-Shot Learning with Class Imbalance

标题：班级不平衡情况下的少机会学习

作者：Mateusz Ochal,Massimiliano Patacchiola,Amos Storkey,Jose Vazquez,Sen Wang
机构：University of Edinburgh, UK, SeeByte Ltd., Edinburgh, UK
备注：[Under Review]
链接：arxiv.org/abs/2101.02523

机器翻译，仅供参考
点击阅读原文访问www.arxivdaily.com,获取带摘要及更多学科的学术速递。

文章推荐