{"id":19320159,"url":"https://github.com/52cv/cvprw-2021-papers","last_synced_at":"2026-03-01T15:34:13.465Z","repository":{"id":111173539,"uuid":"359385571","full_name":"52CV/CVPRW-2021-Papers","owner":"52CV","description":null,"archived":false,"fork":false,"pushed_at":"2021-06-08T02:04:28.000Z","size":221,"stargazers_count":12,"open_issues_count":0,"forks_count":0,"subscribers_count":2,"default_branch":"main","last_synced_at":"2025-11-16T15:22:22.928Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":null,"language":null,"has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/52CV.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null}},"created_at":"2021-04-19T08:33:07.000Z","updated_at":"2023-12-23T14:34:28.000Z","dependencies_parsed_at":"2023-05-06T22:47:28.033Z","dependency_job_id":null,"html_url":"https://github.com/52CV/CVPRW-2021-Papers","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/52CV/CVPRW-2021-Papers","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/52CV%2FCVPRW-2021-Papers","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/52CV%2FCVPRW-2021-Papers/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/52CV%2FCVPRW-2021-Papers/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/52CV%2FCVPRW-2021-Papers/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/52CV","download_url":"https://codeload.github.com/52CV/CVPRW-2021-Papers/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/52CV%2FCVPRW-2021-Papers/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":29973320,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-03-01T15:29:09.406Z","status":"ssl_error","status_checked_at":"2026-03-01T15:28:28.558Z","response_time":124,"last_error":"SSL_connect returned=1 errno=0 peeraddr=140.82.121.5:443 state=error: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":[],"created_at":"2024-11-10T01:27:20.359Z","updated_at":"2026-03-01T15:34:13.428Z","avatar_url":"https://github.com/52CV.png","language":null,"funding_links":[],"categories":[],"sub_categories":[],"readme":"# CVPR2021 Workshop 组会最新论文/代码(持续更新)\n\n## :star2:[CVPR2021最新信息及已接收论文/代码(持续更新)](https://github.com/52CV/CVPR-2021-Papers)\n\n### :fireworks::fireworks::fireworks:更新提示：6月8日新增2篇\n* [Reducing the feature divergence of RGB and near-infrared images using Switchable Normalization](https://arxiv.org/abs/2106.03088)\n* 元学习\n  * [DAMSL: Domain Agnostic Meta Score-based Learning](https://arxiv.org/abs/2106.03041)\n\n### :fireworks::fireworks::fireworks:更新提示：6月7日新增1篇\n* 细粒度\n  * [Fine-Grained Visual Classification of Plant Species In The Wild: Object Detection as A Reinforced Means of Attention](https://arxiv.org/abs/2106.02141)\n\n### :fireworks::fireworks::fireworks:更新提示：6月4日新增2篇\n* Transformer\n  * [Anticipative Video Transformer](https://arxiv.org/abs/2106.02036)\u003cbr\u003e:house:[project](https://facebookresearch.github.io/AVT/)\u003cbr\u003e在 CVPR 21 EPIC-Kitchens 行动预期挑战排行榜上排名第一\n* 图像处理\n  * [NTIRE 2021 Challenge on High Dynamic Range Imaging: Dataset, Methods and Results](https://arxiv.org/abs/2106.01439)\n\n### :fireworks::fireworks::fireworks:更新提示：6月3日新增3篇\n* 半监督\n  * [The Semi-Supervised iNaturalist Challenge at the FGVC8 Workshop](https://arxiv.org/abs/2106.01364)\n* 分割\n  * [Rethinking Cross-modal Interaction from a Top-down Perspective for Referring Video Object Segmentation](https://arxiv.org/abs/2106.01061)\n* 细粒度\n  * [Cleaning and Structuring the Label Space of the iMet Collection 2020](https://arxiv.org/abs/2106.00815)\u003cbr\u003e:star:[code](https://github.com/sunniesuhyoung/iMet2020cleaned)\n\n### :fireworks::fireworks::fireworks:更新提示：6月2日新增3篇\n* 语义分割\n  * [Detecting Anomalies in Semantic Segmentation with Prototypes](https://arxiv.org/abs/2106.00472)\n* 未分\n  * [PanoDR: Spherical Panorama Diminished Reality for Indoor Scenes](https://arxiv.org/abs/2106.00446)\n  * [Semi-Supervised Disparity Estimation with Deep Feature Reconstruction](https://arxiv.org/abs/2106.00318)\n\n### :fireworks::fireworks::fireworks:更新提示：6月1日新增3篇\n* 车辆\n  * [Connecting Language and Vision for Natural Language-Based Vehicle Retrieval](https://arxiv.org/abs/2105.14897)\u003cbr\u003e:star:[code](https://github.com/ShuaiBai623/AIC2021-T5-CLV)\n* 目标检测\n  * [Training Domain-invariant Object Detector Faster with Feature Replay and Slow Learner](https://arxiv.org/abs/2105.14693)\u003cbr\u003e:star:[code](https://github.com/2-Chae/A-NDFT)\u003cbr\u003e本文所介绍算法 A-NDFT，是对 NDFT 的改良版本。A-NDFT 利用两种加速技术，feature replay 和 slow learner。因此，在一个大规模的 UAVDT 基准上，它可以将 NDVT 的训练时间从 31 小时减少到 3 小时，同时仍然保持性能。\n* 6D\n  * [Data-driven 6D Pose Tracking by Calibrating Image Residuals in Synthetic Domains](https://arxiv.org/abs/2105.14391)\u003cbr\u003e:star:[code](https://github.com/wenbowen123/iros20-6d-pose-tracking)\n\n### :fireworks::fireworks::fireworks:更新提示：5月31日新增1篇\n* [The Herbarium 2021 Half–Earth Challenge Dataset](https://arxiv.org/abs/2105.13808)\n\n### :fireworks::fireworks::fireworks:更新提示：5月28日新增1篇\n* [RSCA: Real-time Segmentation-based Context-Aware Scene Text Detection](https://arxiv.org/abs/2105.12789)\n\n### :fireworks::fireworks::fireworks:更新提示：5月27日新增1篇\n* 计算成像\n  * [How to Calibrate Your Event Camera](https://arxiv.org/abs/2105.12362)\u003cbr\u003e:star:[code](https://github.com/uzh-rpg/e2calib)\n\n### :fireworks::fireworks::fireworks:更新提示：5月26日新增1篇\n* 三维\n  * [Real-time Monocular Depth Estimation with Sparse Supervision on Mobile](https://arxiv.org/abs/2105.12053)\n\n\n|:cat:|:dog:|:mouse:|:hamster:|:tiger:|\n|------|------|------|------|------|\n|[38.Transformer](#38)|[37.6D](#37)|[36.OCR](#36)|\n|[35.Data Augmentation(数据增广)](#35)|[34.Computational Photography(光学、几何、光场成像、计算摄影)](#34)|[33.GAN](#33)|[32.手语识别](#32)|[31.图像分类](#31)|\n|[30.目标跟踪](#30)|[29.Auto-ML\u0026NAS](#29)|[28.医学影像](#28)|[27.人体姿态估计](#27)|[26.无监督](#26)|\n|[25.SLAM/AR/VR/机器人](#25)|[24.模型压缩\u0026应用部署](#24)|[23.人脸](#23)|[22.重建](#22)|[21.视频](#21)|\n|[20.三维](#20)|[19.光流](#19)|[18.图像检索](#18)|[17.动作检测识别](#17)|[16.人员重识别](#16)|\n|[15.遥感航空影像](#15)|[14VQA](#14)|[13.SR](#13)|[12.图像分割](#12)|[11.图像处理](#11)|\n|[10.目标检测](#10)|[9.姿态估计](#9)|[8.Camera Trap Images-相机陷阱图像](#8)|[7.图像到图像翻译](#7)|[6.手绘草图](#6)|\n|[5.车辆车牌与智能驾驶](#5)|[4.数据集](#4)|[3.各种神经网络](#3)|[2.算法学习](#2)|[1.Unkown(未分)](#1)|\n\n\u003ca name=\"38\"/\u003e\n\n## 38.Transformer\n  * [Anticipative Video Transformer](https://arxiv.org/abs/2106.02036)\u003cbr\u003e:house:[project](https://facebookresearch.github.io/AVT/)\u003cbr\u003e在 CVPR 21 EPIC-Kitchens 行动预期挑战排行榜上排名第一\n\n\u003ca name=\"37\"/\u003e\n\n## 37.6D\n  * [Data-driven 6D Pose Tracking by Calibrating Image Residuals in Synthetic Domains](https://arxiv.org/abs/2105.14391)\u003cbr\u003e:star:[code](https://github.com/wenbowen123/iros20-6d-pose-tracking)\n\n\u003ca name=\"36\"/\u003e\n\n## 36.OCR\n* 场景文本识别\n  * [RSCA: Real-time Segmentation-based Context-Aware Scene Text Detection](https://arxiv.org/abs/2105.12789)\n\n\u003ca name=\"35\"/\u003e\n\n## 35.Data Augmentation(数据增广)\n  * [Wisdom for the Crowd: Discoursive Power in Annotation Instructions for Computer Vision](https://arxiv.org/abs/2105.10990)\n\n\u003ca name=\"34\"/\u003e\n\n## 34.Computational Photography(光学、几何、光场成像、计算摄影)\n* [How to Calibrate Your Event Camera](https://arxiv.org/abs/2105.12362)\u003cbr\u003e:star:[code](https://github.com/uzh-rpg/e2calib)\n* HDR 成像\n  * [ADNet: Attention-guided Deformable Convolutional Network for High Dynamic Range Imaging](https://arxiv.org/abs/2105.10697)\u003cbr\u003e:star:[code](https://github.com/Pea-Shooter/ADNet)\n  \n\u003ca name=\"33\"/\u003e\n\n## 33.GAN\n* [Learning to Generate Novel Scene Compositions from Single Images and Videos](https://arxiv.org/abs/2105.05847)\n* [Directional GAN: A Novel Conditioning Strategy for Generative Networks](https://arxiv.org/abs/2105.05712)\n\n\u003ca name=\"32\"/\u003e\n\n## 32.手语识别\n* [ChaLearn LAP Large Scale Signer Independent Isolated Sign Language Recognition Challenge: Design, Results and Future Research](https://arxiv.org/abs/2105.05066)\n\n\u003ca name=\"31\"/\u003e\n\n## 31.图像分类\n  * [Boosting Co-teaching with Compression Regularization for Label Noise](https://arxiv.org/abs/2104.13766)\u003cbr\u003e:star:[code](https://github.com/yingyichen-cyy/Nested-Co-teaching)\n* 多标签分类\n  * [PLM: Partial Label Masking for Imbalanced Multi-label Classification](https://arxiv.org/abs/2105.10782)\n* 细粒度\n  * [Cleaning and Structuring the Label Space of the iMet Collection 2020](https://arxiv.org/abs/2106.00815)\u003cbr\u003e:star:[code](https://github.com/sunniesuhyoung/iMet2020cleaned)\n  * [Fine-Grained Visual Classification of Plant Species In The Wild: Object Detection as A Reinforced Means of Attention](https://arxiv.org/abs/2106.02141)\n\n\u003ca name=\"30\"/\u003e\n\n## 30.目标跟踪\n* [Detecting and Matching Related Objects with One Proposal Multiple Predictions](https://arxiv.org/abs/2104.12574)\n* [Differentiable Event Stream Simulator for Non-Rigid 3D Tracking](https://arxiv.org/abs/2104.15139)\u003cbr\u003e:house:[project](http://gvv.mpi-inf.mpg.de/projects/Event-based_Non-rigid_3D_Tracking/)\n* [City-Scale Multi-Camera Vehicle Tracking Guided by Crossroad Zones](https://arxiv.org/abs/2105.06623)\u003cbr\u003e:star:[code](https://github.com/LCFractal/AIC21-MTMC)\n\n\n\u003ca name=\"29\"/\u003e\n\n## 29.Auto-ML\u0026NAS\n* Auto-ML\n  * [Network Space Search for Pareto-Efficient Spaces](https://arxiv.org/abs/2104.11014)\n\n\u003ca name=\"28\"/\u003e\n\n## 28.Medical Imaging医学影像\n* [GAN-Based Data Augmentation and Anonymization for Skin-Lesion Analysis: A Critical Review](https://arxiv.org/abs/2104.10603)\n* 医学图像识别\n  * [Can self-training identify suspicious ugly duckling lesions?](https://arxiv.org/abs/2105.07116)\n* 无监督检测\n  * [Unsupervised Detection of Cancerous Regions in Histology Imagery using Image-to-Image Translation](https://arxiv.org/abs/2104.13786)\n\n\u003ca name=\"27\"/\u003e\n\n## 27.人体姿态估计\n  * [Table Tennis Stroke Recognition Using Two-Dimensional Human Pose Estimation](https://arxiv.org/abs/2104.09907)\n\n\u003ca name=\"26\"/\u003e\n\n## 26.无监督/半监督\n* 无监督\n  * [Perceptual Loss for Robust Unsupervised Homography Estimation](https://arxiv.org/abs/2104.10011)\n* 半监督\n  * [The Semi-Supervised iNaturalist Challenge at the FGVC8 Workshop](https://arxiv.org/abs/2106.01364)\n\n\u003ca name=\"25\"/\u003e\n\n## 25.SLAM/AR/VR/机器人\n  * [Comparing Representations in Tracking for Event Camera-based SLAM](https://arxiv.org/abs/2104.09887)\u003cbr\u003e:star:[code](https://github.com/gogojjh/ESVO_extension)\n\n\u003ca name=\"24\"/\u003e\n\n## 24.Quantization/Pruning/Knowledge Distillation/Model Compression(量化、剪枝、蒸馏、模型压缩/扩展与优化)\n* [BasisNet: Two-stage Model Synthesis for Efficient Inference](https://arxiv.org/abs/2105.03014)\n* 知识蒸馏\n  * [Distill on the Go: Online knowledge distillation in self-supervised learning](https://arxiv.org/abs/2104.09866)\n* 量化\n  * [Do All MobileNets Quantize Poorly? Gaining Insights into the Effect of Quantization on Depthwise Separable Convolutional Networks Through the Eyes of Multi-scale Distributional Dynamics](https://arxiv.org/abs/2104.11849)\n  * [Pareto-Optimal Quantized ResNet Is Mostly 4-bit](https://arxiv.org/abs/2105.03536)\u003cbr\u003e:star:[code](https://github.com/google-research/google-research/tree/master/aqt)\n\n\u003ca name=\"23\"/\u003e\n\n## 23.Face人脸\n* 人脸表情识别\n  * [I Only Have Eyes for You: The Impact of Masks On Convolutional-Based Facial Expression Recognition](https://arxiv.org/abs/2104.08353)\n* 人脸识别\n  * [EQFace: A Simple Explicit Quality Network for Face Recognition](https://arxiv.org/abs/2105.00634)\u003cbr\u003e:star:[code](https://github.com/deepcam-cn/facequality)\n\n\u003ca name=\"22\"/\u003e\n\n## 22.Reconstruction重建\n* 3D 人体重建\n  * [Temporal Consistency Loss for High Resolution Textured and Clothed 3DHuman Reconstruction from Monocular Video](https://arxiv.org/abs/2104.09259)\n\n\u003ca name=\"21\"/\u003e\n\n## 21.Video视频\n* 视频恢复\n  * [Restoration of Video Frames from a Single Blurred Image with Motion Understanding](https://arxiv.org/abs/2104.09134)\n* 异常检测\n  * [An Efficient Approach for Anomaly Detection in Traffic Videos](https://arxiv.org/abs/2104.09758)\n  * [Good Practices and A Strong Baseline for Traffic Anomaly Detection](https://arxiv.org/abs/2105.03827)\u003cbr\u003e在 CVPR 2021 NVIDIA AI CITY 挑战赛中的 Traffic Anomaly Detection(交通异常检测)中排名第一\n* 风格迁移\n  * [Automatic Non-Linear Video Editing Transfer](https://arxiv.org/abs/2105.06988)\n\n\u003ca name=\"20\"/\u003e\n\n## 20.3D三维\n* [OmniLayout: Room Layout Reconstruction from Indoor Spherical Panoramas](https://arxiv.org/abs/2104.09403)\u003cbr\u003e:star:[code](https://github.com/rshivansh/OmniLayout)\n* 深度估计\n  * [Real-time Monocular Depth Estimation with Sparse Supervision on Mobile](https://arxiv.org/abs/2105.12053)\n\n\u003ca name=\"19\"/\u003e\n\n## 19.Optical Flow光流\n* [OmniFlow: Human Omnidirectional Optical Flow](https://arxiv.org/abs/2104.07960)\u003cbr\u003e:sunflower:[dataset](https://www.tu-chemnitz.de/etit/dst/forschung/comp_vision/datasets/omniflow/)\n\n\u003ca name=\"18\"/\u003e\n\n## 18.Image Retrieval图像检索\n* [Continual learning in cross-modal retrieval](https://arxiv.org/abs/2104.06806)\n\n\u003ca name=\"17\"/\u003e\n\n## 17.Action Detection and Recognition动作检测识别\n* action spotting-重点动作识别\n  * [Temporally-Aware Feature Pooling for Action Spotting in Soccer Broadcasts](https://arxiv.org/abs/2104.06779)\n  * [Camera Calibration and Player Localization in SoccerNet-v2 and Investigation of their Representations for Action Spotting](https://arxiv.org/abs/2104.09333)\u003cbr\u003e:sunflower:[dataset](https://soccer-net.org/)\n* 动作检测\n  * [Three-stream network for enriched Action Recognition](https://arxiv.org/abs/2104.13051)\n\n\u003ca name=\"16\"/\u003e\n\n## 16.Person Re-Identifications人员重识别\n* [Graph-based Person Signature for Person Re-Identifications](https://arxiv.org/abs/2104.06770)\n* 行人检测\n  * [Generalizable Multi-Camera 3D Pedestrian Detection](https://arxiv.org/abs/2104.05813)\n* 基于视频的 Reid\n  * [Video-based Person Re-identification without Bells and Whistles](https://arxiv.org/abs/2105.10678)\u003cbr\u003e:star:[code](https://github.com/jackie840129/CF-AAN)\n\n\u003ca name=\"15\"/\u003e\n\n## 15.Aeria/Drones/Satellite/RS Image(航空影像/无人机)\n* 三维重建\n  * [Machine-learned 3D Building Vectorization from Satellite Imagery](https://arxiv.org/abs/2104.06485)\n\n\u003ca name=\"14\"/\u003e\n\n## 14VQA-视觉问答\n* [Dealing with Missing Modalities in the Visual Question Answer-Difference Prediction Task through Knowledge Distillation](https://arxiv.org/abs/2104.05965)\n\n\u003ca name=\"13\"/\u003e\n\n## 13.SR-超分辨率\n* 视频超分辨率\n  * [Efficient Space-time Video Super Resolution using Low-Resolution Flow and Mask Upsampling](https://arxiv.org/abs/2104.05778)\n  * [NTIRE 2021 Challenge on Video Super-Resolution](https://arxiv.org/abs/2104.14852)\u003cbr\u003e:house:[project](https://data.vision.ee.ethz.ch/cvl/ntire21/)\n* 图像超分辨率\n  * [Anchor-based Plain Net for Mobile Image Super-Resolution](https://arxiv.org/abs/2105.09750)\u003cbr\u003e:star:[code](https://github.com/NJU-Jet/SR_Mobile_Quantization)\n\n\u003ca name=\"12\"/\u003e\n\n## 12.Image Segmentation图像分割\n* 语义分割\n  * [Improving Online Performance Prediction for Semantic Segmentation](https://arxiv.org/abs/2104.05255) \n  * [Plants Don't Walk on the Street: Common-Sense Reasoning for Reliable Semantic Segmentation](https://arxiv.org/abs/2104.09254)\n  * [Rethinking Ensemble-Distillation for Semantic Segmentation Based Unsupervised Domain Adaptation](https://arxiv.org/abs/2104.14203)\n  * [Detecting Anomalies in Semantic Segmentation with Prototypes](https://arxiv.org/abs/2106.00472)\n* 实例分割\n  * [Fashion-Guided Adversarial Attack on Person Segmentation](https://arxiv.org/abs/2104.08422)\n* 视频目标分割\n  * [Rethinking Cross-modal Interaction from a Top-down Perspective for Referring Video Object Segmentation](https://arxiv.org/abs/2106.01061)\n\n\u003ca name=\"11\"/\u003e\n\n## 11.Image Processing图像处理\n* [NTIRE 2021 Challenge on High Dynamic Range Imaging: Dataset, Methods and Results](https://arxiv.org/abs/2106.01439)\n* 去除滤镜\n  * [Instagram Filter Removal on Fashionable Images](https://arxiv.org/abs/2104.05072)\n* 去雾\n  * [A Two-branch Neural Network for Non-homogeneous Dehazing via Ensemble Learning](https://arxiv.org/abs/2104.08902)\u003cbr\u003e:star:[code](https://github.com/liuh127/Two-branch-dehazing)\n* 图像压缩\n  * [DANICE: Domain adaptation without forgetting in neural image compression](https://arxiv.org/abs/2104.09370) \n* 图像质量评估\n  * [Region-Adaptive Deformable Network for Image Quality Assessment](https://arxiv.org/abs/2104.11599)\u003cbr\u003e:star:[code](https://github.com/IIGROUP/RADN)\n  * [Perceptual Image Quality Assessment with Transformers](https://arxiv.org/abs/2104.14730)\u003cbr\u003e:star:[code](https://github.com/manricheon/IQT)\u003cbr\u003e在NTIRE 2021年感知IQA挑战中获得第一名\n* 去雨\n  * [Multi-Scale Hourglass Hierarchical Fusion Network for Single Image Deraining](https://arxiv.org/abs/2104.12100)\n* 照片补光\n  * [NTIRE 2021 Depth Guided Image Relighting Challenge](https://arxiv.org/abs/2104.13365)\u003cbr\u003e:star:[code](https://github.com/majedelhelou/VIDIT)\n* 去模糊\n  * [NTIRE 2021 Challenge on Image Deblurring](https://arxiv.org/abs/2104.14854)\u003cbr\u003e:house:[project](https://data.vision.ee.ethz.ch/cvl/ntire21/)\n* 图像补光\n  * [Multi-modal Bifurcated Network for Depth Guided Image Relighting](https://arxiv.org/abs/2105.00690)\u003cbr\u003e:star:[code](https://github.com/weitingchen83/NTIRE2021-Depth-   Guided-Image-Relighting-MBNet)\u003cbr\u003e是 NTIRE 2021 深度指南一对一补光挑战赛的冠军\n  * [S3Net: A Single Stream Structure for Depth Guided Image Relighting](https://arxiv.org/abs/2105.00681)\u003cbr\u003e:star:[code](https://github.com/dectrfov/NTIRE-2021-Depth-Guided-Image-Any-to-Any-relighting)\u003cbr\u003e在 NTIRE 2021 深度引导的任意重新照明挑战中获得第3名\n  * [Physically Inspired Dense Fusion Networks for Relighting](https://arxiv.org/abs/2105.02209)\u003cbr\u003e OIDDR-Net排名第二，AMIDR-Net 在 NTIRE 2021 年深度引导图像重光挑战中名列前五名\n* 图像恢复\n  * [EDPN: Enhanced Deep Pyramid Network for Blurry Image Restoration](https://arxiv.org/abs/2105.04872)\u003cbr\u003e:star:[code](https://github.com/zeyuxiao1997/EDPN)\n* bokeh effect(背景虚化)\n  * [Stacked Deep Multi-Scale Hierarchical Network for Fast Bokeh Effect Rendering from a Single Image](https://arxiv.org/abs/2105.07174)\u003cbr\u003e:star:[code](https://github.com/saikatdutta/Stacked_DMSHN_bokeh)\n\n\u003ca name=\"10\"/\u003e\n\n## 10.Object Detection目标检测\n* [LSPnet: A 2D Localization-oriented Spacecraft Pose Estimation Neural Network](https://arxiv.org/abs/2104.09248)\n* [Pseudo-IoU: Improving Label Assignment in Anchor-Free Object Detection](https://arxiv.org/abs/2104.14082)\u003cbr\u003e:star:[code](https://github.com/SHI-Labs/Pseudo-IoU-for-Anchor-Free-Object-Detection)\n* [Training Domain-invariant Object Detector Faster with Feature Replay and Slow Learner](https://arxiv.org/abs/2105.14693)\u003cbr\u003e:star:[code](https://github.com/2-Chae/A-NDFT)\u003cbr\u003e本文所介绍算法 A-NDFT，是对 NDFT 的改良版本。A-NDFT 利用两种加速技术，feature replay 和 slow learner。因此，在一个大规模的 UAVDT 基准上，它可以将 NDVT 的训练时间从 31 小时减少到 3 小时，同时仍然保持性能。\n* 3D目标检测\n  * [High-level camera-LiDAR fusion for 3D object detection with machine learning](https://arxiv.org/abs/2105.11060)\n\n\n\u003ca name=\"9\"/\u003e\n\n## 9.Pose Estimation姿态估计\n* [Towards Automated and Marker-less Parkinson Disease Assessment: Predicting UPDRS Scores using Sit-stand videos](https://arxiv.org/abs/2104.04650)\n\n\u003ca name=\"8\"/\u003e\n\n## 8.Camera Trap Images-相机陷阱图像\n* [Filtering Empty Camera Trap Images in Embedded Systems](https://arxiv.org/abs/2104.08859)\u003cbr\u003e:star:[code](https://github.com/alcunha/filtering-empty-camera-trap-images)\n\n\u003ca name=\"7\"/\u003e\n\n## 7.Image-to-Image Translation图像到图像翻译\n* [Dual Contrastive Learning for Unsupervised Image-to-Image Translation](https://arxiv.org/abs/2104.07689)\u003cbr\u003e:star:[code](https://github.com/JunlinHan/DCLGAN)\n\n\u003ca name=\"6\"/\u003e\n\n## 6.手绘草图\n* [On Training Sketch Recognizers for New Domains](https://arxiv.org/abs/2104.08850)\n* 工程草图生成\n  * [Engineering Sketch Generation for Computer-Aided Design](https://arxiv.org/abs/2104.09621)\n\n\u003ca name=\"5\"/\u003e\n\n## 5.车辆车牌与智能驾驶\n* 自动驾驶\n  * [MVFuseNet: Improving End-to-End Object Detection and Motion Forecasting through Multi-View Fusion of LiDAR Data](https://arxiv.org/abs/2104.10772)\n  * [Multi-task Learning with Attention for End-to-end Autonomous Driving](https://arxiv.org/abs/2104.10753)\n  * [End-to-End Interactive Prediction and Planning with Optical Flow Distillation for Autonomous Driving](https://arxiv.org/abs/2104.08862)\u003cbr\u003e:house:[project](https://sites.google.com/view/inmp-ofd)\n  * [Rethinking of Radar's Role: A Camera-Radar Dataset and Systematic Annotator via Coordinate Alignment](https://arxiv.org/abs/2105.05207)\u003cbr\u003e:sunflower:[dataset](https://www.cruwdataset.org/)\n* 车辆重识别\n  * [A Strong Baseline for Vehicle Re-Identification](https://arxiv.org/abs/2104.10850)\u003cbr\u003e:star:[code](https://github.com/cybercore-co-ltd/track2_aicity_2021)\n  * [An Empirical Study of Vehicle Re-Identification on the AI City Challenge](https://arxiv.org/abs/2105.09701)\u003cbr\u003e:star:[code](https://github.com/michuanhaohao/AICITY2021_Track2_DMT)\u003cbr\u003e获得 CVPR 2021研讨会上，NVIDIA AI City Challenge（英伟达人工智能城市挑战赛）第2赛道（车辆重识别）的第一名。\n\n* 车辆检索\n  * [SBNet: Segmentation-based Network for Natural Language-based Vehicle Search](https://arxiv.org/abs/2104.11589)\u003cbr\u003e:star:[code](https://github.com/lsrock1/nlp_search)\n  * [Connecting Language and Vision for Natural Language-Based Vehicle Retrieval](https://arxiv.org/abs/2105.14897)\u003cbr\u003e:star:[code](https://github.com/ShuaiBai623/AIC2021-T5-CLV)\n\n\n\u003ca name=\"4\"/\u003e\n\n## 4.Dataset数据集\n* [The Multi-Agent Behavior Dataset: Mouse Dyadic Social Interactions](https://arxiv.org/abs/2104.02710)\n* [The iWildCam 2021 Competition Dataset](https://arxiv.org/abs/2105.03494)\n* [GOO: A Dataset for Gaze Object Prediction in Retail Environments](https://arxiv.org/abs/2105.10793)\u003cbr\u003e:sunflower:[dataset](https://github.com/upeee/GOO-GAZE2021)\n* [The Herbarium 2021 Half–Earth Challenge Dataset](https://arxiv.org/abs/2105.13808)\n\n\u003ca name=\"3\"/\u003e\n\n## 3.各种神经网络\n* CNN\n  * [AsymmNet: Towards ultralight convolution neural networks using asymmetrical bottlenecks](https://arxiv.org/abs/2104.07770)\u003cbr\u003e:star:[code](https://github.com/Spark001/AsymmNet)\n* DNN\n  * [Fast Walsh-Hadamard Transform and Smooth-Thresholding Based Binary Layers in Deep Neural Networks](https://arxiv.org/abs/2104.07085)     \n* BNN-二进制神经网络\n  * [A Bop and Beyond: A Second Order Optimizer for Binarized Neural Networks](https://arxiv.org/abs/2104.05124)\n\n\u003ca name=\"2\"/\u003e\n\n## 2.算法学习\n* 主动学习\n  * [A Mathematical Analysis of Learning Loss for Active Learning in Regression](https://arxiv.org/abs/2104.09315)\n* 对比学习\n  * [Contrastive Learning Improves Model Robustness Under Label Noise](https://arxiv.org/abs/2104.08984)\n* 类增量学习\n  * [Class-Incremental Learning with Generative Classifiers](https://arxiv.org/abs/2104.10093)\u003cbr\u003e:star:[code](https://github.com/GMvandeVen/class-incremental-learning)\n* 增量学习\n  * [IB-DRR: Incremental Learning with Information-Back Discrete Representation Replay](https://arxiv.org/abs/2104.10588)\n* 持续学习\n  * [Class-Incremental Experience Replay for Continual Learning under Concept Drift](https://arxiv.org/abs/2104.11861)\n* 联邦学习\n  * [Towards Fair Federated Learning with Zero-Shot Data Augmentation](https://arxiv.org/abs/2104.13417)\n  * [Cluster-driven Graph Federated Learning over Multiple Domains](https://arxiv.org/abs/2104.14628)\n* 元学习\n  * [DAMSL: Domain Agnostic Meta Score-based Learning](https://arxiv.org/abs/2106.03041)\n\n\n\u003ca name=\"1\"/\u003e\n\n## 1.Unkown未分\n* [Reconsidering CO2 emissions from Computer Vision](https://arxiv.org/abs/2104.08702)\n* [Assessment of deep learning based blood pressure prediction from PPG and rPPG signals](https://arxiv.org/abs/2104.09313)\n* [I Find Your Lack of Uncertainty in Computer Vision Disturbing](https://arxiv.org/abs/2104.08188)\n* [Revisiting The Evaluation of Class Activation Mapping for Explainability: A Novel Metric and Experimental Analysis](https://arxiv.org/abs/2104.10252)\n* [Patch Shortcuts: Interpretable Proxy Models Efficiently Find Black-Box Vulnerabilities](https://arxiv.org/abs/2104.11691)\n* [The 5th AI City Challenge](https://arxiv.org/abs/2104.12233)\n* [Width Transfer: On the (In)variance of Width Optimization](https://arxiv.org/abs/2104.13255)\n* [Sign Segmentation with Changepoint-Modulated Pseudo-Labelling](https://arxiv.org/abs/2104.13817)\n* [CASSOD-Net: Cascaded and Separable Structures of Dilated Convolution for Embedded Vision Systems and Applications](https://arxiv.org/abs/2104.14126)\n* [Feedback control of event cameras](https://arxiv.org/abs/2105.00409)\n* [Effectively Leveraging Attributes for Visual Similarity](https://arxiv.org/abs/2105.01695)\n* [Dynamic-OFA: Runtime DNN Architecture Switching for Performance Scaling on Heterogeneous Embedded Platforms](https://arxiv.org/abs/2105.03596)\u003cbr\u003e与最先进的技术相比，在Jetson Xavier NX 上使用 ImageNet 的实验结果表明，在相似的 ImageNet Top-1 精度下，该方法的速度最高可达 3.5倍（CPU），2.4倍（GPU），或者在相似的延迟下，精度更高 3.8%（CPU），5.1%（GPU）。\n* [High-Resolution Complex Scene Synthesis with Transformers](https://arxiv.org/pdf/2105.06458.pdf)\n* [Deep Graphics Encoder for Real-Time Video Makeup Synthesis from Example](https://arxiv.org/abs/2105.06407)\n* [Texture Generation with Neural Cellular Automata](https://arxiv.org/abs/2105.07299)\u003cbr\u003e:house:[project](https://selforglive.github.io/cvpr_textures/)\n* [Single View Geocentric Pose in the Wild](https://arxiv.org/abs/2105.08229)\u003cbr\u003e:star:[code](https://github.com/pubgeo/monocular-geocentric-pose)\n* [PAL: Intelligence Augmentation using Egocentric Visual Context Detection](https://arxiv.org/abs/2105.10735)\n* [PanoDR: Spherical Panorama Diminished Reality for Indoor Scenes](https://arxiv.org/abs/2106.00446)\n* [Semi-Supervised Disparity Estimation with Deep Feature Reconstruction](https://arxiv.org/abs/2106.00318)\n* [Reducing the feature divergence of RGB and near-infrared images using Switchable Normalization](https://arxiv.org/abs/2106.03088)\n* 异常检测\n  * [Brittle Features May Help Anomaly Detection](https://arxiv.org/abs/2104.10453)\n\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2F52cv%2Fcvprw-2021-papers","html_url":"https://awesome.ecosyste.ms/projects/github.com%2F52cv%2Fcvprw-2021-papers","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2F52cv%2Fcvprw-2021-papers/lists"}