ICCV2021-Papers-with-Code-Demo

☪️论文下载：

ICCV2021 论文下载汇总：

链接: https://pan.baidu.com/s/1vmOQzLG1QaBCgQD1ijtYuw

提取码: bp9j （解压密码，联系微信 nvshenj125 获取）

CVPR 2021整理：https://github.com/DWCTOD/CVPR2021-Papers-with-Code-Demo

论文下载：https://pan.baidu.com/share/init?surl=gjfUQlPf73MCk4vM8VbzoA

密码：aicv

🌟 ICCV 2021持续更新最新论文/paper和相应的开源代码/code！

🚗 ICCV 2021 收录列表

🚂ICCV 2021 报告和demo视频汇总 https://space.bilibili.com/288489574

🚗 官网链接：http://iccv2021.thecvf.com/home

⏲️ 时间 ⌚ 论文/paper接收公布时间：2021年7月23日

✋ ​注：欢迎各位大佬提交issue，分享ICCV 2021论文/paper和开源项目！共同完善这个项目

✈️ 为了方便下载，已将论文/paper存储在文件夹中 ✔️ 表示论文/paper已下载 / Paper Download

🎆 欢迎进群 | Welcome

ICCV 2021 论文/paper交流群已成立！已经收录的同学，可以添加微信：nvshenj125，请备注：ICCV+姓名+学校/公司名称！一定要根据格式申请，可以拉你进群。

image

🔨 目录 |Table of Contents（点击直接跳转）

• Backbone
• Dataset
• Loss
• NAS
• Image Classification
• Vision Transformer
• 目标检测/Object Detection
• 显著性检测/Salient Object Detection
• 3D目标检测 / 3D Object Detection
• 目标跟踪 / Object Tracking
• Image Semantic Segmentation
• Semantic Scene Segmentation
• 3D Semantic Segmentation
• 3D Instance Segmentation
• 实例分割/Instance Segmentation
• 视频分割 / video semantic segmentation
• 医学图像分割/ Medical Image Segmentation
• 医学图像分析/Medical Image Analysis
• GAN
• Style Transfer
• 细粒度分类/Fine-Grained Visual Categorization
• Multi-Label Recognition
• Long-Tailed Recognition
• Geometric deep learning
• Zero/Few Shot
• Unsupervised
• Self-supervised
• Semi Supervised
• Weakly Supervised
• Active Learning
• Action Detection
• 动作识别/Action Recognition
• 时序行为检测 / Temporal Action Localization
• 手语识别/Sign Language Recognition
• Hand Pose Estimation
• Pose Estimation
• 6D Object Pose Estimation
• Human Reconstruction
• 3D Scene Understanding
• 人脸识别/Face Recognition
• 人脸对齐/Face Alignment
• 人脸编辑/Facial Editing
• Face Reconstruction
• Facial Expression Recognition
• 行人重识别/Re-Identification
• Vehicle Re-identification
• Pedestrian Detection
• 人群计数 / Crowd Counting
• Motion Forecasting
• Pedestrian Trajectory Prediction
• Face-Anti-spoofing
• deepfake
• 对抗攻击/ Adversarial Attacks
• 跨模态检索/Cross-Modal Retrieval
• 神经辐射场/NeRF
• 阴影去除/Shadow Removal
• 图像检索/Image Retrieval
• 超分辨/Super-Resolution
• 图像重建/Image Reconstruction
• 去模糊/Image Deblurring
• 去噪/Image Denoising
• 去雪/Image Desnowing
• 图像增强/Image Enhancement
• 图像匹配/Image Matching
• 图像质量/Image Quality
• 图像压缩/Image Compression
• 图像复原/Image Restoration
• 图像修复/Image Inpainting
• 视频修复/Video Inpainting
• 视频插帧/Video Frame Interpolation
• Video Reasoning
• Video Recognition
• Visual Question Answering
• Matching
• 人机交互/Hand-object Interaction
• 视线估计 / Gaze Estimation
• 深度估计 / Depth Estimation
• Contrastive-Learning
• Graph Convolution Networks
• 模型压缩/Compress
• Quantization
• Knowledge Distillation
• 点云/Point Cloud
• 3D reconstruction
• 字体生成/Font Generation
• 文本检测 / Text Detection
• 文本识别 / Text Recognition
• Scene Text Recognizer
• Autonomous-Driving
• Visdrone_detection
• 异常检测 / Anomaly Detection
• 其他/Others

Backbone

✔️Conformer: Local Features Coupling Global Representations for Visual Recognition

• 论文/paper：https://arxiv.org/abs/2105.03889
• 代码/code：https://github.com/pengzhiliang/Conformer

Contextual Convolutional Neural Networks

• 论文/paper：https://arxiv.org/abs/2108.07387
• 代码/code：https://github.com/iduta/coconv

Pyramid Vision Transformer: A Versatile Backbone for Dense Prediction without Convolutions

• 解读：https://zhuanlan.zhihu.com/p/353222035

• 论文/paper：https://arxiv.org/abs/2102.12122

• 代码/code：https://github.com/whai362/PVT

Reg-IBP: Efficient and Scalable Neural Network Robustness Training via Interval Bound Propagation

• 论文/paper：None
• 代码/code：https://github.com/harrywuhust2022/Reg_IBP_ICCV2021

Why Approximate Matrix Square Root Outperforms Accurate SVD in Global Covariance Pooling?

• 论文/paper：https://arxiv.org/abs/2105.02498
• 代码/code：https://github.com/KingJamesSong/DifferentiableSVD

返回目录/back

Dataset

Beyond Road Extraction: A Dataset for Map Update using Aerial Images

• 论文/paper：https://arxiv.org/abs/2110.04690
• 代码/code： None

✔️FineAction: A Fined Video Dataset for Temporal Action Localization

• 论文/paper：https://arxiv.org/abs/2105.11107 | 主页/Homepage
• 代码/code： None

KoDF: A Large-scale Korean DeepFake Detection Dataset

• 论文/paper：https://arxiv.org/abs/2103.10094
• 代码/code：https://moneybrain-research.github.io/kodf

LLVIP: A Visible-infrared Paired Dataset for Low-light Vision

• 论文/paper：https://arxiv.org/abs/2108.10831 | 主页/Homepage
• 代码/code： None

Matching in the Dark: A Dataset for Matching Image Pairs of Low-light Scenes

• 论文/paper：https://arxiv.org/abs/2109.03585
• 代码/code： None

Meta Self-Learning for Multi-Source Domain Adaptation: A Benchmark

• 论文/paper：https://arxiv.org/abs/2108.10840 | 主页/Homepage
• 代码/code：https://github.com/bupt-ai-cz/Meta-SelfLearning

✔️MultiSports: A Multi-Person Video Dataset of Spatio-Temporally Localized Sports Actions

• 论文/paper：https://arxiv.org/abs/2105.07404 | 主页/Homepage
• 代码/code：https://github.com/MCG-NJU/MultiSports/

Semantically Coherent Out-of-Distribution Detection

• 论文/paper：https://arxiv.org/abs/2108.11941 | 主页/Homepage
• 代码/code：https://github.com/jingkang50/ICCV21_SCOOD

StereOBJ-1M: Large-scale Stereo Image Dataset for 6D Object Pose Estimation

• 论文/paper：https://arxiv.org/abs/2109.10115
• 代码/code：None

STRIVE: Scene Text Replacement In Videos

• 论文/paper：https://arxiv.org/abs/2109.02762 | 主页/Homepage
• 代码/code：None

The Many Faces of Robustness: A Critical Analysis of Out-of-Distribution Generalization

• 论文/paper：https://arxiv.org/abs/2006.16241
• 代码/code：https://github.com/hendrycks/imagenet-r

Webly Supervised Fine-Grained Recognition: Benchmark Datasets and An Approach

• 论文/paper：https://arxiv.org/abs/2108.02399
• 代码/code：https://github.com/NUST-Machine-Intelligence-Laboratory/weblyFG-dataset

Who's Waldo? Linking People Across Text and Images (Oral)

• 论文/paper：https://arxiv.org/abs/2108.07253
• 代码/code：None

返回目录/back

Loss

Asymmetric Loss For Multi-Label Classification

• 论文/paper：https://arxiv.org/abs/2009.14119
• 代码/code：https://github.com/Alibaba-MIIL/ASL

Bias Loss for Mobile Neural Networks

• 论文/paper：https://arxiv.org/abs/2107.11170
• 代码/code：None

Focal Frequency Loss for Image Reconstruction and Synthesis

• 论文/paper：https://arxiv.org/abs/2012.12821
• 代码/code：https://github.com/EndlessSora/focal-frequency-loss

Orthogonal Projection Loss

• 论文/paper：https://arxiv.org/abs/2103.14021
• 代码/code：https://github.com/kahnchana/opl

Rank & Sort Loss for Object Detection and Instance Segmentation (Oral)

• 论文/paper：https://arxiv.org/abs/2107.11669
• 代码/code：https://github.com/kemaloksuz/RankSortLoss

返回目录/back

NAS

BN-NAS: Neural Architecture Search with Batch Normalization

• 论文/paper：https://arxiv.org/abs/2108.07375
• 代码/code：None

BossNAS: Exploring Hybrid CNN-transformers with Block-wisely Self-supervised Neural Architecture Search

• 论文/paper：https://arxiv.org/pdf/2103.12424.pdf
• 代码/code：https://github.com/changlin31/BossNAS

CONet: Channel Optimization for Convolutional Neural Networks

• 论文/paper：https://arxiv.org/abs/2108.06822
• 代码/code：None

FOX-NAS: Fast, On-device and Explainable Neural Architecture Search

• 论文/paper：https://arxiv.org/abs/2108.08189
• 代码/code：https://github.com/great8nctu/FOX-NAS

Pi-NAS: Improving Neural Architecture Search by Reducing Supernet Training Consistency Shift

• 论文/paper：https://arxiv.org/abs/2108.09671v1
• 代码/code：https://github.com/Ernie1/Pi-NAS

RANK-NOSH: Efficient Predictor-Based Architecture Search via Non-Uniform Successive Halving

• 论文/paper：https://arxiv.org/abs/2108.08019
• 代码/code：https://github.com/ruocwang

Single-DARTS: Towards Stable Architecture Search

• 论文/paper：https://arxiv.org/abs/2108.08128
• 代码/code：https://github.com/PencilAndBike/Single-DARTS.git

返回目录/back

Image Classification

Influence-Balanced Loss for Imbalanced Visual Classification

• 论文/paper：https://arxiv.org/abs/2110.02444
• 代码/code：None

Low-Shot Validation: Active Importance Sampling for Estimating Classifier Performance on Rare Categories

• 论文/paper：https://arxiv.org/abs/2109.05720
• 代码/code：None

Tune It or Don't Use It: Benchmarking Data-Efficient Image Classification

• 论文/paper：https://arxiv.org/abs/2108.13122
• 代码/code：None

返回目录/back

Vision Transformer

An End-to-End Transformer Model for 3D Object Detection

• 论文/paper：https://arxiv.org/abs/2109.08141
• 代码/code：None

AutoFormer: Searching Transformers for Visual Recognition

• 论文/paper：https://arxiv.org/abs/2107.00651
• 代码/code：https://github.com/microsoft/AutoML

BossNAS: Exploring Hybrid CNN-transformers with Block-wisely Self-supervised Neural Architecture Search

• 论文/paper：https://arxiv.org/pdf/2103.12424.pdf
• 代码/code：https://github.com/changlin31/BossNAS

Conditional DETR for Fast Training Convergence

• 论文/paper：https://arxiv.org/abs/2108.06152
• 代码/code：https://git.io/ConditionalDETR

Dyadformer: A Multi-modal Transformer for Long-Range Modeling of Dyadic Interactions

• 论文/paper：https://arxiv.org/abs/2109.09487
• 代码/code：None

Eformer: Edge Enhancement based Transformer for Medical Image Denoising

• 论文/paper：https://arxiv.org/abs/2109.08044
• 代码/code：None

Fast Convergence of DETR with Spatially Modulated Co-Attention

• 解读：https://zhuanlan.zhihu.com/p/397083124
• 论文/paper：https://arxiv.org/abs/2108.02404
• 代码/code：https://github.com/gaopengcuhk/SMCA-DETR

FuseFormer: Fusing Fine-Grained Information in Transformers for Video Inpainting

• 论文/paper：https://arxiv.org/abs/2108.01912
• 代码/code：None

Generic Attention-model Explainability for Interpreting Bi-Modal and Encoder-Decoder Transformers (Oral)

• 论文/paper：https://arxiv.org/pdf/2103.15679.pdf
• 代码/code：https://github.com/hila-chefer/Transformer-MM-Explainability

GroupFormer: Group Activity Recognition with Clustered Spatial-Temporal Transformer

• 论文/paper：https://arxiv.org/abs/2108.12630
• 代码/code：https://github.com/xueyee/GroupFormer

HiFT: Hierarchical Feature Transformer for Aerial Tracking

• 论文/paper：https://arxiv.org/abs/2108.00202
• 代码/code：https://github.com/vision4robotics/HiFT

High-Fidelity Pluralistic Image Completion with Transformers

• 论文/paper：https://arxiv.org/pdf/2103.14031.pdf | 主页/Homepage
• 代码/code： https://github.com/raywzy/ICT

Improving 3D Object Detection with Channel-wise Transformer

• 论文/paper：https://arxiv.org/abs/2108.10723
• 代码/code：None

Is it Time to Replace CNNs with Transformers for Medical Images?

• 论文/paper：https://arxiv.org/abs/2108.09038
• 代码/code：None

Learning Spatio-Temporal Transformer for Visual Tracking

• 论文/paper：https://arxiv.org/abs/2103.17154
• 代码/code：https://github.com/researchmm/Stark

MUSIQ: Multi-scale Image Quality Transformer

• 论文/paper：https://arxiv.org/abs/2108.05997
• 代码/code：None

Paint Transformer: Feed Forward Neural Painting with Stroke Prediction (Oral)

• 解读：https://zhuanlan.zhihu.com/p/400017971

• 论文/paper：https://arxiv.org/abs/2108.03798

• 代码/code：https://github.com/Huage001/PaintTransformer

PlaneTR: Structure-Guided Transformers for 3D Plane Recovery

• 论文/paper：https://arxiv.org/abs/2107.13108
• 代码/code： https://github.com/IceTTTb/PlaneTR3D

PnP-DETR: Towards Efficient Visual Analysis with Transformers

• 论文/paper：https://arxiv.org/abs/2109.07036
• 代码/code：https://github.com/twangnh/pnp-detr

Pose Transformers (POTR): Human Motion Prediction with Non-Autoregressive Transformers

• 论文/paper：https://arxiv.org/abs/2109.07531
• 代码/code：None

PoinTr: Diverse Point Cloud Completion with Geometry-Aware Transformers (Oral)

• 论文/paper：https://arxiv.org/abs/2108.08839
• 代码/code：https://github.com/yuxumin/PoinTr

Pyramid Vision Transformer: A Versatile Backbone for Dense Prediction without Convolutions

• 解读：https://zhuanlan.zhihu.com/p/353222035

• 论文/paper：https://arxiv.org/abs/2102.12122

• 代码/code：https://github.com/whai362/PVT

Rethinking and Improving Relative Position Encoding for Vision Transformer

• 论文/paper：https://houwenpeng.com/publications/iRPE.pdf
• 代码/code：https://github.com/wkcn/iRPE-model-zoo

Rethinking Spatial Dimensions of Vision Transformers

• 论文/paper：https://arxiv.org/abs/2103.16302
• 代码/code：https://github.com/naver-ai/pit

Simpler is Better: Few-shot Semantic Segmentation with Classifier Weight Transformer

• 论文/paper：https://arxiv.org/abs/2108.03032
• 代码/code：https://github.com/zhiheLu/CWTfor-FSS

SnowflakeNet: Point Cloud Completion by Snowflake Point Deconvolution with Skip-Transformer

• 论文/paper：https://arxiv.org/abs/2108.04444
• 代码/code：https://github.com/AllenXiangX/SnowflakeNet

Spatial-Temporal Transformer for Dynamic Scene Graph Generation

• 解读：用于视频场景图生成的Spatial-Temporal Transformer
• 论文/paper：https://arxiv.org/abs/2107.12309
• 代码/code：None

SOTR: Segmenting Objects with Transformers

• 论文/paper：https://arxiv.org/abs/2108.06747
• 代码/code：https://github.com/easton-cau/SOTR

Swin Transformer: Hierarchical Vision Transformer using Shifted Windows

• 论文/paper：https://arxiv.org/abs/2103.14030
• 代码/code：https://github.com/microsoft/Swin-Transformer

Revisiting Stereo Depth Estimation From a Sequence-to-Sequence Perspective with Transformers

• 论文/paper：https://arxiv.org/abs/2011.02910
• 代码/code：https://github.com/mli0603/stereo-transformer

The Animation Transformer: Visual Correspondence via Segment Matching

• 论文/paper：https://arxiv.org/abs/2109.02614
• 代码/code：None

The Right to Talk: An Audio-Visual Transformer Approach

• 论文/paper：https://arxiv.org/abs/2108.03256
• 代码/code：None

TPH-YOLOv5: Improved YOLOv5 Based on Transformer Prediction Head for Object Detection on Drone-captured Scenarios

• 论文/paper：https://arxiv.org/abs/2108.11539
• 代码/code：None

TransFER: Learning Relation-aware Facial Expression Representations with Transformers

• 论文/paper：https://arxiv.org/abs/2108.11116
• 代码/code：None

TransPose: Keypoint Localization via Transformer

• 论文/paper：https://arxiv.org/abs/2012.14214
• 代码/code：https://github.com/yangsenius/TransPose

✔️Tokens-to-Token ViT: Training Vision Transformers from Scratch on ImageNet

• 论文/paper：https://arxiv.org/abs/2101.11986

• 代码/code： https://github.com/yitu-opensource/T2T-ViT

✔️Visual Transformer with Statistical Test for COVID-19 Classification

• 论文/paper：https://arxiv.org/abs/2107.05334
• 代码/code： None

Vision Transformer with Progressive Sampling

• 论文/paper：https://arxiv.org/abs/2108.01684
• 代码/code：https://github.com/yuexy/PS-ViT

Visual Saliency Transformer

• 解读：https://blog.csdn.net/qq_39936426/article/details/117199411

• 论文/paper：https://arxiv.org/abs/2104.12099

• 代码/code： https://github.com/nnizhang/VST

Vision-Language Transformer and Query Generation for Referring Segmentation

• 论文/paper：https://arxiv.org/abs/2108.05565
• 代码/code：https://github.com/henghuiding/Vision-Language-Transformer

Voxel Transformer for 3D Object Detection

• 论文/paper：https://arxiv.org/abs/2109.02497
• 代码/code： None

返回目录/back

目标检测/Object Detection

Active Learning for Deep Object Detection via Probabilistic Modeling

• 论文/paper：https://arxiv.org/abs/2103.16130
• 代码/code：None

Boosting Weakly Supervised Object Detection via Learning Bounding Box Adjusters

• 论文/paper：https://arxiv.org/abs/2108.01499
• 代码/code：https://github.com/DongSky/lbba_boosted_wsod

Change is Everywhere: Single-Temporal Supervised Object Change Detection in Remote Sensing Imagery

• 论文/paper：https://arxiv.org/abs/2108.07002
• 代码/code：https://github.com/Z-Zheng/ChangeStar

Conditional Variational Capsule Network for Open Set Recognition

• 论文/paper： https://arxiv.org/abs/2104.09159

• 代码/code：https://github.com/guglielmocamporese/cvaecaposr

DetCo: Unsupervised Contrastive Learning for Object Detection

• 论文/paper：https://arxiv.org/abs/2102.04803
• 代码/code： https://github.com/xieenze/DetCo

DeFRCN: Decoupled Faster R-CNN for Few-Shot Object Detection

• 论文/paper：https://arxiv.org/abs/2108.09017
• 代码/code：None

Deployment of Deep Neural Networks for Object Detection on Edge AI Devices with Runtime Optimization

• 论文/paper：https://arxiv.org/abs/2108.08166
• 代码/code：None

Detecting Invisible People

• 论文/paper：https://arxiv.org/abs/2012.08419 | 主页/Homepage
• 代码/code：None

FMODetect: Robust Detection and Trajectory Estimation of Fast Moving Objects

• 论文/paper：None
• 代码/code：https://github.com/rozumden/FMODetect

GraphFPN: Graph Feature Pyramid Network for Object Detection

• 论文/paper：https://arxiv.org/abs/2108.00580
• 代码/code：None

Human Detection and Segmentation via Multi-view Consensus

• 论文/paper：None
• 代码/code：https://github.com/isinsukatircioglu/mvc

MDETR : Modulated Detection for End-to-End Multi-Modal Understanding

• 论文/paper：https://arxiv.org/abs/2104.12763 | 主页/Homepage
• 代码/code： https://github.com/ashkamath/mdetr

Mutual Supervision for Dense Object Detection

• 论文/paper：https://arxiv.org/abs/2109.05986
• 代码/code：None

Morphable Detector for Object Detection on Demand

• 论文/paper：https://arxiv.org/abs/2110.04917
• 代码/code：https://github.com/Zhaoxiangyun/Morphable-Detector

Moving Object Detection for Event-based vision using Graph Spectral Clustering

• 论文/paper：https://arxiv.org/abs/2109.14979
• 代码/code：None

Oriented R-CNN for Object Detection

• 论文/paper：https://arxiv.org/abs/2108.05699
• 代码/code：https://github.com/jbwang1997/OBBDetection

Rank & Sort Loss for Object Detection and Instance Segmentation (Oral)

• 论文/paper：https://arxiv.org/abs/2107.11669
• 代码/code：https://github.com/kemaloksuz/RankSortLoss

Reconcile Prediction Consistency for Balanced Object Detection

• 论文/paper：https://arxiv.org/abs/2108.10809
• 代码/code：None

Seeking Similarities over Differences: Similarity-based Domain Alignment for Adaptive Object Detection

• 论文/paper：https://arxiv.org/abs/2110.01428
• 代码/code：None

Towards Rotation Invariance in Object Detection

• 论文/paper：https://arxiv.org/abs/2109.13488
• 代码/code：None

TOOD: Task-aligned One-stage Object Detection (Oral)

• 论文/paper：https://arxiv.org/abs/2108.07755
• 代码/code：https://github.com/fcjian/TOOD

Vector-Decomposed Disentanglement for Domain-Invariant Object Detection

• 论文/paper：https://arxiv.org/abs/2108.06685
• 代码/code：None

返回目录/back

Salient Object Detections

Disentangled High Quality Salient Object Detection

• 论文/paper：https://arxiv.org/abs/2108.03551
• 代码/code：None

Light Field Saliency Detection with Dual Local Graph Learning andReciprocative Guidance

• 论文/paper：https://arxiv.org/abs/2110.00698
• 代码/code：None

RGB-D Saliency Detection via Cascaded Mutual Information Minimization

• 论文/paper：https://arxiv.org/abs/2109.07246
• 代码/code：https://github.com/JingZhang617/cascaded_rgbd_sod

Specificity-preserving RGB-D Saliency Detection

• 论文/paper：https://arxiv.org/abs/2108.08162
• 代码/code：https://github.com/taozh2017/SPNet

Summarize and Search: Learning Consensus-aware Dynamic Convolution for Co-Saliency Detection

• 论文/paper：https://arxiv.org/abs/2109.07246
• 代码/code：https://github.com/nnizhang/CADC

返回目录/back

3D目标检测 / 3D Object Detection

An End-to-End Transformer Model for 3D Object Detection

• 论文/paper：https://arxiv.org/abs/2109.08141
• 代码/code：None

Fog Simulation on Real LiDAR Point Clouds for 3D Object Detection in Adverse Weather

• 论文/paper：https://arxiv.org/abs/2108.05249
• 代码/code：https://github.com/MartinHahner/LiDAR_fog_sim

LIGA-Stereo: Learning LiDAR Geometry Aware Representations for Stereo-based 3D Detector

• 论文/paper：https://arxiv.org/abs/2108.08258
• 代码/code：None

MonoCInIS: Camera Independent Monocular 3D Object Detection using Instance Segmentation

• 论文/paper：https://arxiv.org/abs/2110.00464
• 代码/code：None

Improving 3D Object Detection with Channel-wise Transformer

• 论文/paper：https://arxiv.org/abs/2108.10723
• 代码/code：None

Is Pseudo-Lidar needed for Monocular 3D Object detection?

• 论文/paper：https://arxiv.org/abs/2108.06417
• 代码/code：None

ODAM: Object Detection, Association, and Mapping using Posed RGB Video （Oral）

• 论文/paper：https://arxiv.org/abs/2108.10165v1
• 代码/code：None

Pyramid R-CNN: Towards Better Performance and Adaptability for 3D Object Detection

• 论文/paper：https://arxiv.org/abs/2109.02499
• 代码/code：None

RandomRooms: Unsupervised Pre-training from Synthetic Shapes and Randomized Layouts for 3D Object Detection

• 论文/paper：https://arxiv.org/abs/2108.07794
• 代码/code：None

Voxel Transformer for 3D Object Detection

• 论文/paper：https://arxiv.org/abs/2109.02497
• 代码/code： None

Unsupervised Domain Adaptive 3D Detection with Multi-Level Consistency

• 论文/paper：https://arxiv.org/pdf/2107.11355.pdf
• 代码/code：None

返回目录/back

目标跟踪 / Object Tracking

DepthTrack : Unveiling the Power of RGBD Tracking

• 论文/paper：https://arxiv.org/abs/2108.13962
• 代码/code：None

Exploring Simple 3D Multi-Object Tracking for Autonomous Driving

• 论文/paper：https://arxiv.org/abs/2108.10312v1
• 代码/code：None

Is First Person Vision Challenging for Object Tracking?

• 论文/paper：https://arxiv.org/abs/2108.13665
• 代码/code：None

Learning to Track Objects from Unlabeled Videos

• 论文/paper：https://arxiv.org/abs/2108.12711
• 代码/code：https://github.com/VISION-SJTU/USOT

Learn to Match: Automatic Matching Network Design for Visual Tracking

• 论文/paper：https://arxiv.org/abs/2108.00803
• 代码/code：https://github.com/JudasDie/SOTS

Making Higher Order MOT Scalable: An Efficient Approximate Solver for Lifted Disjoint Paths

• 论文/paper：https://arxiv.org/abs/2108.10606
• 代码/code：https://github.com/TimoK93/ApLift

Saliency-Associated Object Tracking

• 论文/paper：https://arxiv.org/abs/2108.03637
• 代码/code：None

Video Annotation for Visual Tracking via Selection and Refinement

• 论文/paper：https://arxiv.org/abs/2108.03821
• 代码/code：https://github.com/Daikenan/VASR

返回目录/back

Image Semantic Segmentation

Complementary Patch for Weakly Supervised Semantic Segmentation

• 论文/paper：https://arxiv.org/abs/2108.03852
• 代码/code：None

Calibrated Adversarial Refinement for Stochastic Semantic Segmentation

• 论文/paper：https://arxiv.org/abs/2006.13144
• 代码/code：https://github.com/EliasKassapis/CARSSS

Deep Metric Learning for Open World Semantic Segmentation

• 论文/paper：https://arxiv.org/abs/2108.04562
• 代码/code：None

Dual Path Learning for Domain Adaptation of Semantic Segmentation

• 论文/paper：https://arxiv.org/abs/2108.06337
• 代码/code：https://github.com/royee182/DPL

EdgeFlow: Achieving Practical Interactive Segmentation with Edge-Guided Flow

• 论文/paper：https://arxiv.org/abs/2109.09406
• 代码/code：https://github.com/PaddlePaddle/PaddleSeg

Exploiting Spatial-Temporal Semantic Consistency for Video Scene Parsing

• 论文/paper：https://arxiv.org/abs/2109.02281
• 代码/code：None

Exploiting a Joint Embedding Space for Generalized Zero-Shot Semantic Segmentation

• 论文/paper：https://arxiv.org/abs/2108.06536
• 代码/code：None

Exploring Cross-Image Pixel Contrast for Semantic Segmentation （Oral）

• 论文/paper：https://arxiv.org/abs/2101.11939
• 代码/code：https://github.com/tfzhou/ContrastiveSeg

Enhanced Boundary Learning for Glass-like Object Segmentation

• 论文/paper：https://arxiv.org/abs/2103.15734
• 代码/code：https://github.com/hehao13/EBLNet

From Contexts to Locality: Ultra-high Resolution Ie Segmentation via Locality-aware Contextual Correlation

• 论文/paper：https://arxiv.org/abs/2109.02580
• 代码/code：https://github.com/liqiokkk/FCtL

ISNet: Integrate Image-Level and Semantic-Level Context for Semantic Segmentation

• 论文/paper：https://arxiv.org/abs/2108.12382v1
• 代码/code：None

Generalize then Adapt: Source-Free Domain Adaptive Semantic Segmentation

• 论文/paper：https://arxiv.org/abs/2108.11249
• 代码/code：https://sites.google.com/view/sfdaseg

Labels4Free: Unsupervised Segmentation using StyleGAN

• 论文/paper：https://arxiv.org/abs/2103.14968 | 主页/Homepage
• 代码/code：None

LabOR: Labeling Only if Required for Domain Adaptive Semantic Segmentation

• 论文/paper：https://arxiv.org/abs/2108.05570
• 代码/code：None

Learning Meta-class Memory for Few-Shot Semantic Segmentation

• 论文/paper：https://arxiv.org/abs/2108.02958
• 代码/code：None

Leveraging Auxiliary Tasks with Affinity Learning for Weakly Supervised Semantic Segmentation

• 论文/paper：https://arxiv.org/abs/2107.11787
• 代码/code：https://github.com/xulianuwa/AuxSegNet

Mining Contextual Information Beyond Image for Semantic Segmentation

• 论文/paper：https://arxiv.org/abs/2108.11819
• 代码/code：None

Mining Latent Classes for Few-shot Segmentation(Oral)

• 论文/paper：https://arxiv.org/abs/2103.15402
• 代码/code：https://github.com/LiheYoung/MiningFSS

Multi-Target Adversarial Frameworks for Domain Adaptation in Semantic Segmentation

• 论文/paper：https://arxiv.org/abs/2108.06962
• 代码/code：None

Multi-Anchor Active Domain Adaptation for Semantic Segmentation (Oral)

• 论文/paper：https://arxiv.org/abs/2108.08012
• 代码/code：None

Personalized Image Semantic Segmentation

• 论文/paper：None
• 代码/code： https://github.com/zhangyuygss/PIS

Pixel Contrastive-Consistent Semi-Supervised Semantic Segmentation

• 论文/paper：https://arxiv.org/abs/2108.09025
• 代码/code：None

Pseudo-mask Matters inWeakly-supervised Semantic Segmentation

• 论文/paper：https://arxiv.org/abs/2108.12995
• 代码/code：https://github.com/Eli-YiLi/PMM

RECALL: Replay-based Continual Learning in Semantic Segmentation

• 论文/paper：https://arxiv.org/abs/2108.03673
• 代码/code：None

Re-distributing Biased Pseudo Labels for Semi-supervised Semantic Segmentation: A Baseline Investigation(Oral)

• 论文/paper：https://arxiv.org/abs/2107.11279
• 代码/code：https://github.com/CVMI-Lab/DARS

Semantic Segmentation on VSPW Dataset through Aggregation of Transformer Models

• 论文/paper：https://arxiv.org/abs/2109.01316
• 代码/code：None

Self-Regulation for Semantic Segmentation

• 论文/paper：https://arxiv.org/abs/2108.09702v1
• 代码/code：None

Semantic Concentration for Domain Adaptation

• 论文/paper：https://arxiv.org/abs/2108.05720
• 代码/code：None

ShapeConv: Shape-aware Convolutional Layer for Indoor RGB-D Semantic Segmentation

• 论文/paper：https://arxiv.org/abs/2108.10528
• 代码/code：None

Simpler is Better: Few-shot Semantic Segmentation with Classifier Weight Transformer

• 论文/paper：https://arxiv.org/abs/2108.03032
• 代码/code：https://github.com/zhiheLu/CWTfor-FSS

SOTR: Segmenting Objects with Transformers

• 论文/paper：https://arxiv.org/abs/2108.06747
• 代码/code：https://github.com/easton-cau/SOTR

Standardized Max Logits: A Simple yet Effective Approach for Identifying Unexpected Road Obstacles in Urban-Scene Segmentation

• 论文/paper：https://arxiv.org/abs/2107.11264v1
• 代码/code：None

The Marine Debris Dataset for Forward-Looking Sonar Semantic Segmentation

• 论文/paper：https://arxiv.org/abs/2108.06800
• 代码/code：https://github.com/mvaldenegro/marine-debris-fls-datasets/

Unsupervised Semantic Segmentation by Contrasting Object Mask Proposals

• 论文/paper：https://arxiv.org/pdf/2102.06191.pdf
• 代码/code：https://github.com/wvangansbeke/Unsupervised-Semantic-Segmentation

Weakly Supervised Temporal Anomaly Segmentation with Dynamic Time Warping

• 论文/paper：https://arxiv.org/abs/2108.06816
• 代码/code：None

返回目录/back

Semantic Scene Segmentation

BiMaL: Bijective Maximum Likelihood Approach to Domain Adaptation in Semantic Scene Segmentation

• 论文/paper：https://arxiv.org/abs/2108.03267
• 代码/code：None

返回目录/back

3D Semantic Segmentation

VMNet: Voxel-Mesh Network for Geodesic-aware 3D Semantic Segmentation

• 论文/paper：None
• 代码/code：https://github.com/hzykent/VMNet

返回目录/back

3D Instance Segmentation

Hierarchical Aggregation for 3D Instance Segmentation

• 论文/paper：https://arxiv.org/abs/2108.02350
• 代码/code：https://github.com/hustvl/HAIS

Instance Segmentation in 3D Scenes using Semantic Superpoint Tree Networks

• 论文/paper：https://arxiv.org/abs/2108.07478
• 代码/code：https://github.com/Gorilla-Lab-SCUT/SSTNet

返回目录/back

实例分割/Instance Segmentation

CDNet: Centripetal Direction Network for Nuclear Instance Segmentation

• 论文/paper：None

• 代码/code： https://github.com/2021-ICCV/CDNet

✔️Crossover Learning for Fast Online Video Instance Segmentation

• 论文/paper：https://arxiv.org/abs/2104.05970

• 代码/code： https://github.com/hustvl/CrossVIS

✔️Instances as Queries

• 论文/paper：https://arxiv.org/abs/2105.01928
• 代码/code： https://github.com/hustvl/QueryInst

Instance Segmentation Challenge Track Technical Report, VIPriors Workshop at ICCV 2021: Task-Specific Copy-Paste Data Augmentation Method for Instance Segmentation

• 论文/paper：https://arxiv.org/abs/2110.00470
• 代码/code：https://github.com/jahongir7174/VIP2021

Rank & Sort Loss for Object Detection and Instance Segmentation (Oral)

• 论文/paper：https://arxiv.org/abs/2107.11669
• 代码/code：https://github.com/kemaloksuz/RankSortLoss

Scaling up instance annotation via label propagation

• 论文/paper：https://arxiv.org/abs/2110.02277
• 代码/code：http://scaling-anno.csail.mit.edu/

返回目录/back

视频分割 / video semantic segmentation

Domain Adaptive Video Segmentation via Temporal Consistency Regularization

• 论文/paper：https://arxiv.org/abs/2107.11004
• 代码/code：https://github.com/Dayan-Guan/DA-VSN

Full-Duplex Strategy for Video Object Segmentation

• 论文/paper：https://arxiv.org/abs/2108.03151 | 主页/homepage
• 代码/code：https://github.com/GewelsJI/FSNet

Hierarchical Memory Matching Network for Video Object Segmentation

• demo：https://www.bilibili.com/video/BV1Eg41157q3

• 论文/paper：https://arxiv.org/abs/2109.11404 | 主页/homepage

• 代码/code：Hierarchical Memory Matching Network for Video Object Segmentation

Joint Inductive and Transductive Learning for Video Object Segmentation

• 论文/paper：https://arxiv.org/abs/2108.03679
• 代码/code：https://github.com/maoyunyao/JOINT

返回目录/back

Medical Image Segmentation

Recurrent Mask Refinement for Few-Shot Medical Image Segmentation

• 论文/paper：https://arxiv.org/abs/2108.00622
• 代码/code：None

Uncertainty-aware GAN with Adaptive Loss for Robust MRI Image Enhancement

• 论文/paper：https://arxiv.org/abs/2110.03343
• 代码/code：None

返回目录/back

Medical Image Analysis

Eformer: Edge Enhancement based Transformer for Medical Image Denoising

• 论文/paper：https://arxiv.org/abs/2109.08044
• 代码/code：None

Improving Tuberculosis (TB) Prediction using Synthetically Generated Computed Tomography (CT) Images

• 论文/paper：https://arxiv.org/abs/2109.11480
• 代码/code：None

Preservational Learning Improves Self-supervised Medical Image Models by Reconstructing Diverse Contexts

• 论文/paper：https://arxiv.org/abs/2109.04379
• 代码/code：https://github.com/Luchixiang/PCRL

Studying the Effects of Self-Attention for Medical Image Analysis

• 论文/paper：https://arxiv.org/abs/2109.01486
• 代码/code：None

返回目录/back

GAN

3DStyleNet: Creating 3D Shapes with Geometric and Texture Style Variations (Oral)

• 论文/paper：https://arxiv.org/abs/2108.12958
• 代码/code：https://nv-tlabs.github.io/3DStyleNet/

AdaAttN: Revisit Attention Mechanism in Arbitrary Neural Style Transfer

• 论文/paper：https://arxiv.org/abs/2108.03647
• 代码/code：https://github.com/Huage001/AdaAttN

Click to Move: Controlling Video Generation with Sparse Motion

• 论文/paper：https://arxiv.org/abs/2108.08815
• 代码/code：https://github.com/PierfrancescoArdino/C2M

Collaging Class-specific GANs for Semantic Image Synthesis

• 论文/paper：https://arxiv.org/abs/2110.04281
• 代码/code：None

Disentangled Lifespan Face Synthesis

• 论文/paper：https://arxiv.org/abs/2108.02874 | 主页/Homepage
• 代码/code：https://github.com/SenHe/DLFS

Dual Projection Generative Adversarial Networks for Conditional Image Generation

• 论文/paper：https://arxiv.org/abs/2108.09016
• 代码/code：None

EigenGAN: Layer-Wise Eigen-Learning for GANs

• 论文/paper：https://arxiv.org/pdf/2104.12476.pdf
• 代码/code：https://github.com/LynnHo/EigenGAN-Tensorflow

GAN Inversion for Out-of-Range Images with Geometric Transformations

• 论文/paper：https://arxiv.org/abs/2108.08998
• 代码/code：https://kkang831.github.io/publication/ICCV_2021_BDInvert/

Generative Models for Multi-Illumination Color Constancy

• 论文/paper：https://arxiv.org/abs/2109.00863
• 代码/code：None

Gradient Normalization for Generative Adversarial Networks

• 论文/paper：https://arxiv.org/abs/2109.02235
• 代码/code：None

Graph-to-3D: End-to-End Generation and Manipulation of 3D Scenes Using Scene Graphs

• 论文/paper：https://arxiv.org/abs/2108.08841
• 代码/code：None

Image Synthesis via Semantic Composition

• 论文/paper：https://arxiv.org/abs/2109.07053 | 主页/Homepage
• 代码/code：https://github.com/dvlab-research/SCGAN

InSeGAN: A Generative Approach to Segmenting Identical Instances in Depth Images

• 论文/paper：https://arxiv.org/abs/2108.13865
• 代码/code：None

Learning to Diversify for Single Domain Generalization

• 论文/paper：https://arxiv.org/abs/2108.11726
• 代码/code：None

Manifold Matching via Deep Metric Learning for Generative Modeling

• 论文/paper：https://arxiv.org/abs/2106.10777
• 代码/code：https://github.com/dzld00/pytorch-manifold-matching

Meta Gradient Adversarial Attack

• 论文/paper：https://arxiv.org/abs/2108.04204
• 代码/code：None

Online Multi-Granularity Distillation for GAN Compression

• 论文/paper：https://arxiv.org/abs/2108.06908
• 代码/code：https://github.com/bytedance/OMGD

Orthogonal Jacobian Regularization for Unsupervised Disentanglement in Image Generation

• 论文/paper：https://arxiv.org/abs/2108.07668
• 代码/code：https://github.com/csyxwei/OroJaR

PixelSynth: Generating a 3D-Consistent Experience from a Single Image

• 论文/paper：https://arxiv.org/abs/2108.05892 | 主页/Homepage
• 代码/code：https://github.com/crockwell/pixelsynth/

Robustness and Generalization via Generative Adversarial Training

• 论文/paper：https://arxiv.org/abs/2109.02765
• 代码/code：None

SemIE: Semantically-Aware Image Extrapolation

• 论文/paper：https://arxiv.org/abs/2108.13702
• 代码/code：https://semie-iccv.github.io/

SketchLattice: Latticed Representation for Sketch Manipulation

• 论文/paper：https://arxiv.org/abs/2108.11636
• 代码/code：None

Sketch Your Own GAN

• 论文/paper：https://arxiv.org/abs/2108.02774
• 代码/code：https://github.com/PeterWang512/GANSketching

Target Adaptive Context Aggregation for Video Scene Graph Generation

• 论文/paper：https://arxiv.org/abs/2108.08121
• 代码/code：https://github.com/MCG-NJU/TRACE

Toward a Visual Concept Vocabulary for GAN Latent Space

• 论文/paper：https://arxiv.org/abs/2110.04292
• 代码/code：None

Toward Spatially Unbiased Generative Models

• 论文/paper：https://arxiv.org/abs/2108.01285
• 代码/code：None

Towards Vivid and Diverse Image Colorization with Generative Color Prior

• 论文/paper：https://arxiv.org/abs/2108.08826
• 代码/code：None

Bridging the Gap between Label- and Reference-based Synthesis in Multi-attribute Image-to-Image Translation

• 论文/paper：https://arxiv.org/abs/2110.05055
• 代码/code：None

Unaligned Image-to-Image Translation by Learning to Reweight

• 论文/paper：https://arxiv.org/abs/2109.11736
• 代码/code：None

Unconditional Scene Graph Generation

• 论文/paper：https://arxiv.org/abs/2108.05884
• 代码/code：None

Unsupervised Geodesic-preserved Generative Adversarial Networks for Unconstrained 3D Pose Transfer

• 论文/paper：https://arxiv.org/abs/2108.07520
• 代码/code：https://github.com/mikecheninoulu/Unsupervised_IEPGAN

返回目录/back

Style Transfer

Domain-Aware Universal Style Transfer

• 论文/paper：https://arxiv.org/abs/2108.04441
• 代码/code：None

返回目录/back

细粒度分类/Fine-Grained Visual Categorization

Benchmark Platform for Ultra-Fine-Grained Visual Categorization BeyondHuman Performance

• 论文/paper：None
• 代码/code：https://github.com/XiaohanYu-GU/Ultra-FGVC

Webly Supervised Fine-Grained Recognition: Benchmark Datasets and An Approach

• 论文/paper：https://arxiv.org/abs/2108.02399
• 代码/code：https://github.com/NUST-Machine-Intelligence-Laboratory/weblyFG-dataset

返回目录/back

Multi-Label Recognition

Residual Attention: A Simple but Effective Method for Multi-Label Recognition

• 论文/paper：https://arxiv.org/abs/2108.02456
• 代码/code：None

返回目录/back

Long-Tailed Recognition

ACE: Ally Complementary Experts for Solving Long-Tailed Recognition in One-Shot Oral

• 论文/paper：https://arxiv.org/abs/2108.02385
• 代码/code：https://github.com/jrcai/ACE

返回目录/back

Geometric deep learning

Manifold Matching via Deep Metric Learning for Generative Modeling

• 论文/paper：https://arxiv.org/abs/2106.10777
• 代码/code：https://github.com/dzld00/pytorch-manifold-matching

Orthogonal Jacobian Regularization for Unsupervised Disentanglement in Image Generation

• 论文/paper：None
• 代码/code：https://github.com/csyxwei/OroJaR

返回目录/back

Zero/Few Shot

Binocular Mutual Learning for Improving Few-shot Classification

• 论文/paper：https://arxiv.org/abs/2108.12104v1
• 代码/code：None

Boosting the Generalization Capability in Cross-Domain Few-shot Learning via Noise-enhanced Supervised Autoencoder

• 论文/paper：https://arxiv.org/abs/2108.05028
• 代码/code：None

Discriminative Region-based Multi-Label Zero-Shot Learning

• 论文/paper：https://arxiv.org/abs/2108.05028
• 代码/code：None

Domain Generalization via Gradient Surgery

• 论文/paper：https://arxiv.org/abs/2108.01621
• 代码/code：None

Exploiting a Joint Embedding Space for Generalized Zero-Shot Semantic Segmentation

• 论文/paper：https://arxiv.org/abs/2108.06536
• 代码/code：None

Few-Shot Batch Incremental Road Object Detection via Detector Fusion

• 论文/paper：https://arxiv.org/abs/2108.08048
• 代码/code：None

Field-Guide-Inspired Zero-Shot Learning

• 论文/paper：https://arxiv.org/abs/2108.10967
• 代码/code：None

Few-shot Visual Relationship Co-localization

• 论文/paper：https://arxiv.org/abs/2108.11618
• 代码/code：None

Generalized Source-free Domain Adaptation

• 论文/paper：https://arxiv.org/abs/2108.01614
• 代码/code：https://github.com/Albert0147/G-SFDA

Generalized and Incremental Few-Shot Learning by Explicit Learning and Calibration without Forgetting

• 论文/paper：https://arxiv.org/abs/2108.08165
• 代码/code：None

Meta-Learning with Task-Adaptive Loss Function for Few-Shot Learning

• 论文/paper：https://arxiv.org/abs/2110.03909
• 代码/code：https://github.com/baiksung/MeTAL

Meta Navigator: Search for a Good Adaptation Policy for Few-shot Learning

• 论文/paper：https://arxiv.org/abs/2109.05749
• 代码/code：None

On the Importance of Distractors for Few-Shot Classification

• 论文/paper：https://arxiv.org/abs/2109.09883
• 代码/code：None

Relational Embedding for Few-Shot Classification

• 论文/paper：https://arxiv.org/abs/2108.09666v1
• 代码/code：None

SIGN: Spatial-information Incorporated Generative Network for Generalized Zero-shot Semantic Segmentation

• 论文/paper：https://arxiv.org/abs/2108.12517
• 代码/code：None

Transductive Few-Shot Classification on the Oblique Manifold

• 论文/paper：https://arxiv.org/abs/2108.04009
• 代码/code：None

Visual Domain Adaptation for Monocular Depth Estimation on Resource-Constrained Hardware

• 论文/paper：https://arxiv.org/abs/2108.02671
• 代码/code：None

返回目录/back

Unsupervised

Adversarial Robustness for Unsupervised Domain Adaptation

• 论文/paper：https://arxiv.org/abs/2109.00946
• 代码/code：None

Collaborative Unsupervised Visual Representation Learning from Decentralized Data

• 论文/paper：https://arxiv.org/abs/2108.06492
• 代码/code：None

Instance Similarity Learning for Unsupervised Feature Representation

• 论文/paper：https://arxiv.org/abs/2108.02721
• 代码/code：https://github.com/ZiweiWangTHU/ISL

Skeleton Cloud Colorization for Unsupervised 3D Action Representation Learning

• 论文/paper：https://arxiv.org/abs/2108.01959
• 代码/code：None

Unsupervised Dense Deformation Embedding Network for Template-Free Shape Correspondence

• 论文/paper：https://arxiv.org/abs/2108.11609
• 代码/code：None

Tune it the Right Way: Unsupervised Validation of Domain Adaptation via Soft Neighborhood Density

• 论文/paper：https://arxiv.org/abs/2108.10860
• 代码/code：https://github.com/VisionLearningGroup/SND

返回目录/back

Self-supervised

Digging into Uncertainty in Self-supervised Multi-view Stereo

• 论文/paper：https://arxiv.org/abs/2108.12966
• 代码/code：None

Enhancing Self-supervised Video Representation Learning via Multi-level Feature Optimization

• 论文/paper：https://arxiv.org/abs/2108.02183
• 代码/code：None

Focus on the Positives: Self-Supervised Learning for Biodiversity Monitoring

• 论文/paper：https://arxiv.org/abs/2108.06435
• 代码/code：None

Improving Self-supervised Learning with Hardness-aware Dynamic Curriculum Learning: An Application to Digital Pathology

• 论文/paper：https://arxiv.org/abs/2108.07183
• 代码/code：https://github.com/srinidhiPY/ICCVCDPATH2021-ID-8

Meta Self-Learning for Multi-Source Domain Adaptation: A Benchmark

• 论文/paper：https://arxiv.org/abs/2108.10840 | 主页/Homepage
• 代码/code：https://github.com/bupt-ai-cz/Meta-SelfLearning

Reducing Label Effort: Self-Supervised meets Active Learning

• 论文/paper：https://arxiv.org/abs/2108.11458
• 代码/code：None

Self-supervised Neural Networks for Spectral Snapshot Compressive Imaging

• 论文/paper：https://arxiv.org/abs/2108.12654
• 代码/code：https://github.com/mengziyi64/CASSI-Self-Supervised

Self-Supervised Visual Representations Learning by Contrastive Mask Prediction

• 论文/paper：https://arxiv.org/abs/2108.07954
• 代码/code：None

Self-Supervised Video Representation Learning with Meta-Contrastive Network

• 论文/paper：https://arxiv.org/abs/2108.08426
• 代码/code：None

SSH: A Self-Supervised Framework for Image Harmonization

• 论文/paper：https://arxiv.org/abs/2108.06805
• 代码/code：https://github.com/VITA-Group/SSHarmonization

返回目录/back

Semi Supervised

Trash to Treasure: Harvesting OOD Data with Cross-Modal Matching for Open-Set Semi-Supervised Learning

• 论文/paper：https://arxiv.org/abs/2108.05617
• 代码/code：None

Warp-Refine Propagation: Semi-Supervised Auto-labeling via Cycle-consistency

• 论文/paper：https://arxiv.org/abs/2109.13432
• 代码/code：None

返回目录/back

Weakly Supervised

A Weakly Supervised Amodal Segmenter with Boundary Uncertainty Estimation

• 论文/paper：https://arxiv.org/abs/2108.09897v1
• 代码/code：None

Foreground-Action Consistency Network for Weakly Supervised Temporal Action Localization

• 论文/paper：https://arxiv.org/abs/2108.06524
• 代码/code：https://github.com/LeonHLJ/FAC-Net

Online Refinement of Low-level Feature Based Activation Map for Weakly Supervised Object Localization

• 论文/paper：https://arxiv.org/abs/2110.05741
• 代码/code：None

返回目录/back

Active Learning

Influence Selection for Active Learning

• 论文/paper：https://arxiv.org/abs/2108.09331v1
• 代码/code：None

返回目录/back

Action Detection

Class Semantics-based Attention for Action Detection

• 论文/paper：https://arxiv.org/abs/2109.02613
• 代码/code：None

返回目录/back

Action Recognition

"Knights": First Place Submission for VIPriors21 Action Recognition Challenge at ICCV 2021

• 论文/paper：https://arxiv.org/abs/2110.07758
• 代码/code：None https://vipriors.github.io/challenges/#action-recognition

A Baseline Framework for Part-level Action Parsing and Action Recognition

• 论文/paper：https://arxiv.org/abs/2110.03368
• 代码/code：None

Channel-wise Topology Refinement Graph Convolution for Skeleton-Based Action Recognition

• 论文/paper：https://arxiv.org/abs/2107.12213
• 代码/code：https://github.com/Uason-Chen/CTR-GCN

Elaborative Rehearsal for Zero-shot Action Recognition

• 论文/paper：https://arxiv.org/abs/2108.02833

• 代码/code： https://github.com/DeLightCMU/ElaborativeRehearsal

✔️FineAction: A Fined Video Dataset for Temporal Action Localization

• 论文/paper：https://arxiv.org/abs/2105.11107 | 主页/Homepage

• 代码/code： None

✔️MultiSports: A Multi-Person Video Dataset of Spatio-Temporally Localized Sports Actions

• 论文/paper：https://arxiv.org/abs/2105.07404 | 主页/Homepage
• 代码/code：https://github.com/MCG-NJU/MultiSports/

Spatio-Temporal Dynamic Inference Network for Group Activity Recognition

• 论文/paper：https://arxiv.org/abs/2108.11743
• 代码/code：None

Unsupervised Few-Shot Action Recognition via Action-Appearance Aligned Meta-Adaptation (Oral)

• 论文/paper：https://arxiv.org/abs/2109.15317
• 代码/code：None

Video Pose Distillation for Few-Shot, Fine-Grained Sports Action Recognition

• 论文/paper：https://arxiv.org/abs/2109.01305
• 代码/code：None

返回目录/back

时序行为检测 / Temporal Action Localization

Enriching Local and Global Contexts for Temporal Action Localization

• 论文/paper：https://arxiv.org/abs/2104.02330
• 代码/code：None

Boundary-sensitive Pre-training for Temporal Localization in Videos

• 论文/paper：https://arxiv.org/abs/2011.10830
• 代码/code：None

返回目录/back

手语识别/Sign Language Recognition

SignBERT: Pre-Training of Hand-Model-Aware Representation for Sign Language Recognition

• 论文/paper：https://arxiv.org/abs/2110.05382
• 代码/code：None

Visual Alignment Constraint for Continuous Sign Language Recognition

• 论文/paper：https://arxiv.org/abs/2104.02330
• 代码/code： https://github.com/Blueprintf/VAC_CSLR

返回目录/back

Hand Pose Estimation

HandFoldingNet: A 3D Hand Pose Estimation Network Using Multiscale-Feature Guided Folding of a 2D Hand Skeleton

• 论文/paper：https://arxiv.org/abs/2108.05545
• 代码/code：https://github.com/cwc1260/HandFold

返回目录/back

Pose Estimation

2D Pose Estimation

Hand-Object Contact Consistency Reasoning for Human Grasps Generation

• 论文/paper：https://arxiv.org/pdf/2104.03304.pdf | 主页/Homepage
• 代码/code： None

Human Pose Regression with Residual Log-likelihood Estimation Oral

• 论文/paper：https://arxiv.org/abs/2107.11291| 主页/Homepage
• 代码/code：https://github.com/Jeff-sjtu/res-loglikelihood-regression

Online Knowledge Distillation for Efficient Pose Estimation

• 论文/paper：https://arxiv.org/abs/2108.02092
• 代码/code： None

The Center of Attention: Center-Keypoint Grouping via Attention for Multi-Person Pose Estimation

• 论文/paper：https://arxiv.org/abs/2110.05132
• 代码/code：https://github.com/dvl-tum/center-group

TransPose: Keypoint Localization via Transformer

• 论文/paper：https://arxiv.org/abs/2012.14214
• 代码/code：https://github.com/yangsenius/TransPose

3D Pose Estimation

EventHPE: Event-based 3D Human Pose and Shape Estimation

• 论文/paper：https://arxiv.org/abs/2108.06819
• 代码/code：None

DECA: Deep viewpoint-Equivariant human pose estimation using Capsule Autoencoders（Oral）

• 论文/paper：https://arxiv.org/abs/2108.08557
• 代码/code：https://github.com/mmlab-cv/DECA

FrankMocap: A Monocular 3D Whole-Body Pose Estimation System via Regression and Integration

• 论文/paper：https://arxiv.org/abs/2108.06428
• 代码/code：None

Hierarchical Kinematic Probability Distributions for 3D Human Shape and Pose Estimation from Images in the Wild

• 论文/paper：https://arxiv.org/abs/2110.00990
• 代码/code：https://github.com/akashsengupta1997/HierarchicalProbabilistic3DHuman

Learning Skeletal Graph Neural Networks for Hard 3D Pose Estimation

• 论文/paper：https://arxiv.org/abs/2108.07181
• 代码/code：https://github.com/ailingzengzzz/Skeletal-GNN

Probabilistic-Monocular-3D-Human-Pose-Estimation-with-Normalizing-Flows

• 论文/paper：https://arxiv.org/abs/2107.13788
• 代码/code： https://github.com/twehrbein/Probabilistic-Monocular-3D-Human-Pose-Estimation-with-Normalizing-Flows

PyMAF: 3D Human Pose and Shape Regression with Pyramidal Mesh Alignment Feedback Loop

• 论文/paper：https://arxiv.org/abs/2103.16507 | 主页/Homepage
• 代码/code： https://github.com/HongwenZhang/PyMAF

Shape-aware Multi-Person Pose Estimation from Multi-View Images

• 论文/paper：https://arxiv.org/abs/2110.02330 | 主页/Homepage
• 代码/code：None

Unsupervised 3D Pose Estimation for Hierarchical Dance Video Recognition

• 论文/paper：https://arxiv.org/abs/2109.09166
• 代码/code：None

返回目录/back

6D Object Pose Estimation

RePOSE: Real-Time Iterative Rendering and Refinement for 6D Object Pose Estimation

• 论文/paper：https://arxiv.org/abs/2104.00633
• 代码/code：https://github.com/sh8/RePOSE

SO-Pose: Exploiting Self-Occlusion for Direct 6D Pose Estimation

• 论文/paper：https://arxiv.org/abs/2108.08367
• 代码/code：None

StereOBJ-1M: Large-scale Stereo Image Dataset for 6D Object Pose Estimation

• 论文/paper：https://arxiv.org/abs/2109.10115
• 代码/code：None

返回目录/back

Human Reconstruction

ARCH++: Animation-Ready Clothed Human Reconstruction Revisited

• 论文/paper：https://arxiv.org/abs/2108.07845
• 代码/code：None

imGHUM: Implicit Generative Models of 3D Human Shape and Articulated Pose

• 论文/paper：https://arxiv.org/abs/2108.10842
• 代码/code：None

Learning to Regress Bodies from Images using Differentiable Semantic Rendering

• 论文/paper：https://arxiv.org/abs/2110.03480 | 主页/Homepage
• 代码/code：None

Learning Motion Priors for 4D Human Body Capture in 3D Scenes (Oral)

• 论文/paper：https://arxiv.org/abs/2108.10399 |主页/Homepage
• 代码/code：https://github.com/sanweiliti/LEMO

Physics-based Human Motion Estimation and Synthesis from Videos

• 论文/paper：https://arxiv.org/abs/2109.09913
• 代码/code：None

Probabilistic Modeling for Human Mesh Recovery

• 论文/paper：https://arxiv.org/abs/2108.11944
• 代码/code：https://www.seas.upenn.edu/~nkolot/projects/prohmr/

返回目录/back

3D Scene Understanding

DeepPanoContext: Panoramic 3D Scene Understanding with Holistic Scene Context Graph and Relation-based Optimization (Oral)

• 论文/paper：https://arxiv.org/abs/2108.10743 |主页/Homepage
• 代码/code：None

Estimating and Exploiting the Aleatoric Uncertainty in Surface Normal Estimation [oral]

• 论文/paper：https://arxiv.org/abs/2109.09881
• 代码/code：https://github.com/baegwangbin/surface_normal_uncertainty

返回目录/back

Face Recognition

Masked Face Recognition Challenge: The InsightFace Track Report

• 论文/paper：https://arxiv.org/abs/2108.08191
• 代码/code：https://github.com/deepinsight/insightface/tree/master/challenges/iccv21-mfr

Masked Face Recognition Challenge: The WebFace260M Track Report

• 论文/paper：https://arxiv.org/abs/2108.07189
• 代码/code：None

PASS: Protected Attribute Suppression System for Mitigating Bias in Face Recognition

• 论文/paper：https://arxiv.org/abs/2108.03764
• 代码/code：None

Rethinking Common Assumptions to Mitigate Racial Bias in Face Recognition Datasets

• 论文/paper：https://arxiv.org/abs/2109.03229
• 代码/code：https://github.com/j-alex-hanson/rethinking-race-face-datasets

SynFace: Face Recognition with Synthetic Data

• 论文/paper：https://arxiv.org/abs/2108.07960
• 代码/code：None

Unravelling the Effect of Image Distortions for Biased Prediction of Pre-trained Face Recognition Models

• 论文/paper：https://arxiv.org/abs/2108.06581
• 代码/code：None

返回目录/back

Face Alignment

ADNet: Leveraging Error-Bias Towards Normal Direction in Face Alignment

• 论文/paper：https://arxiv.org/abs/2109.05721
• 代码/code：None

返回目录/back

Facial Editing

Talk-to-Edit: Fine-Grained Facial Editing via Dialog

• 论文/paper：https://arxiv.org/abs/2109.04425 | 主页/Homepage
• 代码/code：https://github.com/yumingj/Talk-to-Edit

返回目录/back

Face Reconstruction

Self-Supervised 3D Face Reconstruction via Conditional Estimation

• 论文/paper：https://arxiv.org/abs/2110.04800

• 代码/code：None

Towards High Fidelity Monocular Face Reconstruction with Rich Reflectance using Self-supervised Learning and Ray Tracing

• 论文/paper：https://arxiv.org/abs/2103.15432

• 代码/code：None

返回目录/back

Facial Expression Recognition

TransFER: Learning Relation-aware Facial Expression Representations with Transformers

• 论文/paper：https://arxiv.org/abs/2108.11116
• 代码/code：None

Understanding and Mitigating Annotation Bias in Facial Expression Recognition

• 论文/paper：https://arxiv.org/abs/2108.08504
• 代码/code：None

返回目录/back

行人重识别/Re-Identification

A Technical Report for ICCV 2021 VIPriors Re-identification Challenge

• 论文/paper：https://arxiv.org/abs/2109.15164
• 代码/code：None

ASMR: Learning Attribute-Based Person Search with Adaptive Semantic Margin Regularizer

• 论文/paper：https://arxiv.org/abs/2108.04533
• 代码/code：None

Counterfactual Attention Learning for Fine-Grained Visual Categorization and Re-identification

• 论文/paper：https://arxiv.org/abs/2108.08728
• 代码/code：https://github.com/raoyongming/CAL

IDM: An Intermediate Domain Module for Domain Adaptive Person Re-ID Oral

• 论文/paper：https://arxiv.org/abs/2108.02413
• 代码/code：https://github.com/SikaStar/IDM

Learning by Aligning: Visible-Infrared Person Re-identification using Cross-Modal Correspondences

• 论文/paper：https://arxiv.org/abs/2108.07422
• 代码/code：None

Learning Instance-level Spatial-Temporal Patterns for Person Re-identification

• 论文/paper：https://arxiv.org/abs/2108.00171

• 代码/code：https://github.com/RenMin1991/cleaned-DukeMTMC-reID/

Learning Compatible Embeddings

• 论文/paper：None
• 代码/code：https://github.com/IrvingMeng/LCE

Multi-Expert Adversarial Attack Detection in Person Re-identification Using Context Inconsistency

• 论文/paper：https://arxiv.org/abs/2108.09891v1
• 代码/code：None

Towards Discriminative Representation Learning for Unsupervised Person Re-identification

• 论文/paper：https://arxiv.org/abs/2108.03439
• 代码/code：None

TransReID: Transformer-based Object Re-Identification

• 论文/paper：https://arxiv.org/abs/2102.04378
• 代码/code：https://github.com/heshuting555/TransReID

Video-based Person Re-identification with Spatial and Temporal Memory Networks

• 论文/paper：https://arxiv.org/abs/2108.09039
• 代码/code：None

Weakly Supervised Person Search with Region Siamese Networks

• 论文/paper：https://arxiv.org/abs/2109.06109
• 代码/code：None

返回目录/back

Vehicle Re-identification

Heterogeneous Relational Complement for Vehicle Re-identification

• 论文/paper：https://arxiv.org/abs/2109.07894
• 代码/code：None

返回目录/back

Pedestrian Detection

MOTSynth: How Can Synthetic Data Help Pedestrian Detection and Tracking?

• 论文/paper：https://arxiv.org/abs/2108.09518v1
• 代码/code：None

Spatial and Semantic Consistency Regularizations for Pedestrian Attribute Recognition

• 论文/paper：https://arxiv.org/abs/2109.05686
• 代码/code：None

返回目录/back

人群计数 /Crowd Counting

Rethinking Counting and Localization in Crowds:A Purely Point-Based Framework (Oral)

• 论文/paper：https://arxiv.org/abs/2107.12746
• 代码/code：https://github.com/TencentYoutuResearch/CrowdCounting-P2PNet

Uniformity in Heterogeneity:Diving Deep into Count Interval Partition for Crowd Counting

• 论文/paper：https://arxiv.org/abs/2107.12619
• 代码/code：https://github.com/TencentYoutuResearch/CrowdCounting-UEPNet

Variational Attention: Propagating Domain-Specific Knowledge for Multi-Domain Learning in Crowd Counting

• 论文/paper：https://arxiv.org/abs/2108.08023
• 代码/code：None

返回目录/back

Motion Forecasting

Generating Smooth Pose Sequences for Diverse Human Motion Prediction

• 论文/paper：https://arxiv.org/abs/2108.08422
• 代码/code：https://github.com/wei-mao-2019/gsps

MSR-GCN: Multi-Scale Residual Graph Convolution Networks for Human Motion Prediction

• 论文/paper：https://arxiv.org/abs/2108.07152
• 代码/code：https://github.com/Droliven/MSRGCN

RAIN: Reinforced Hybrid Attention Inference Network for Motion Forecasting

• 论文/paper：https://arxiv.org/abs/2108.01316 | 主页/Homepage
• 代码/code：None

Skeleton-Graph: Long-Term 3D Motion Prediction From 2D Observations Using Deep Spatio-Temporal Graph CNNs

• 论文/paper：https://arxiv.org/abs/2109.10257
• 代码/code：https://github.com/abduallahmohamed/Skeleton-Graph

返回目录/back

Pedestrian Trajectory Prediction

DenseTNT: End-to-end Trajectory Prediction from Dense Goal Sets

• 论文/paper：https://arxiv.org/abs/2108.09640v1
• 代码/code：None

MG-GAN: A Multi-Generator Model Preventing Out-of-Distribution Samples in Pedestrian Trajectory Prediction

• 论文/paper：https://arxiv.org/abs/2108.09274
• 代码/code：https://github.com/selflein/MG-GAN

返回目录/back

Face-Anti-spoofing

CL-Face-Anti-spoofing

• 论文/paper：None
• 代码/code：https://github.com/xxheyu/CL-Face-Anti-spoofing

3D High-Fidelity Mask Face Presentation Attack Detection Challenge

• 论文/paper：https://arxiv.org/abs/2108.06968
• 代码/code：None

Exploring Temporal Coherence for More General Video Face Forgery Detection

• 论文/paper：https://arxiv.org/abs/2108.06693
• 代码/code：None

返回目录/back

deepfake

OpenForensics: Large-Scale Challenging Dataset For Multi-Face Forgery Detection And Segmentation In-The-Wild

• 论文/paper：https://arxiv.org/abs/2107.14480 | Dataset
• 代码/code：None

Fake It Till You Make It: Face analysis in the wild using synthetic data alone

• 论文/paper：https://arxiv.org/abs/2109.15102
• 代码/code：None

返回目录/back

对抗攻击/ Adversarial Attacks

A Hierarchical Assessment of Adversarial Severity

• 论文/paper：https://arxiv.org/abs/2108.11785
• 代码/code：None

AdvDrop: Adversarial Attack to DNNs by Dropping Information

• 论文/paper：https://arxiv.org/abs/2108.09034
• 代码/code：None

AGKD-BML: Defense Against Adversarial Attack by Attention Guided Knowledge Distillation and Bi-directional Metric Learning

• 论文/paper：https://arxiv.org/abs/2108.06017
• 代码/code：https://github.com/hongw579/AGKD-BML

Optical Adversarial Attack

• 论文/paper：https://arxiv.org/abs/2108.06247
• 代码/code：None

Sample Efficient Detection and Classification of Adversarial Attacks via Self-Supervised Embeddings

• 论文/paper：https://arxiv.org/abs/2108.13797
• 代码/code：None

TkML-AP: Adversarial Attacks to Top-k Multi-Label Learning

• 论文/paper：https://arxiv.org/abs/2108.00146
• 代码/code：None

返回目录/back

跨模态检索/Cross-Modal Retrieval

Wasserstein Coupled Graph Learning for Cross-Modal Retrieval

• 论文/paper：None
• 代码/code：None

返回目录/back

深度估计 / Depth Estimation

AA-RMVSNet: Adaptive Aggregation Recurrent Multi-view Stereo Network

• 论文/paper：https://arxiv.org/abs/2108.03824
• 代码/code：https://github.com/QT-Zhu/AA-RMVSNet

Augmenting Depth Estimation with Geospatial Context

• 论文/paper：https://arxiv.org/abs/2109.09879
• 代码/code：None

Excavating the Potential Capacity of Self-Supervised Monocular Depth Estimation

• 论文/paper：https://arxiv.org/abs/2109.12484
• 代码/code：None

Fine-grained Semantics-aware Representation Enhancement for Self-supervised Monocular Depth Estimation (oral)

• 论文/paper：https://arxiv.org/abs/2108.08829
• 代码/code：https://github.com/hyBlue/FSRE-Depth

Motion Basis Learning for Unsupervised Deep Homography Estimationwith Subspace Projection

• 论文/paper：None
• 代码/code：https://github.com/NianjinYe/Motion-Basis-Homography

Regularizing Nighttime Weirdness: Efficient Self-supervised Monocular Depth Estimation in the Dark

• 论文/paper：https://arxiv.org/abs/2108.03830
• 代码/code：None

Revisiting Stereo Depth Estimation From a Sequence-to-Sequence Perspective with Transformers

• 论文/paper：https://arxiv.org/abs/2011.02910
• 代码/code：https://github.com/mli0603/stereo-transformer

Self-supervised Monocular Depth Estimation for All Day Images using Domain Separation

• 论文/paper：https://arxiv.org/abs/2108.07628
• 代码/code：None

SLIDE: Single Image 3D Photography with Soft Layering and Depth-aware Inpainting (Oral)

• 论文/paper：https://arxiv.org/abs/2109.01068 | 主页/Homepage
• 代码/code：None

StructDepth: Leveraging the structural regularities for self-supervised indoor depth estimation

• 论文/paper：https://arxiv.org/abs/2108.08574
• 代码/code：https://github.com/SJTU-ViSYS/StructDepth

返回目录/back

视频插帧/Video Frame Interpolation

Asymmetric Bilateral Motion Estimation for Video Frame Interpolation

• 论文/paper：https://arxiv.org/abs/2108.06815

• 代码/code：https://github.com/JunHeum/ABME

✔️XVFI: eXtreme Video Frame Interpolation(Oral)

• 论文/paper：https://arxiv.org/abs/2103.16206

• 代码/code： https://github.com/JihyongOh/XVFI

返回目录/back

Video Reasoning

The Multi-Modal Video Reasoning and Analyzing Competition

• 论文/paper：https://arxiv.org/abs/2108.08344
• 代码/code：None

返回目录/back

NeRF

CodeNeRF: Disentangled Neural Radiance Fields for Object Categories

• 论文/paper：https://arxiv.org/abs/2109.01750 | 主页/Homepage
• 代码/code：https://github.com/wayne1123/code-nerf

GNeRF: GAN-based Neural Radiance Field without Posed Camera

• 论文/paper：https://arxiv.org/abs/2103.15606 | 主页/Homepage
• 代码/code：https://github.com/MQ66/gnerf

In-Place Scene Labelling and Understanding with Implicit Scene Representation (Oral)

• 论文/paper：https://arxiv.org/abs/2103.15875 | 主页/Homepage
• 代码/code：None

KiloNeRF: Speeding up Neural Radiance Fields with Thousands of Tiny MLPs

• 论文/paper：https://arxiv.org/abs/2103.13744| 主页/Homepage
• 代码/code：https://github.com/creiser/kilonerf

Learning Object-Compositional Neural Radiance Field for Editable Scene Rendering

• 论文/paper：https://arxiv.org/abs/2109.01847 | 主页/Homepage
• 代码/code：https://github.com/zju3dv/object_nerf

NerfingMVS: Guided Optimization of Neural Radiance Fields for Indoor Multi-view Stereo (Oral)

• 论文/paper：https://arxiv.org/abs/2109.01129 | 主页/Homepage
• 代码/code：https://github.com/weiyithu/NerfingMVS

Putting NeRF on a Diet: Semantically Consistent Few-Shot View Synthesis

• 论文/paper：https://arxiv.org/abs/2104.00677 | 主页/Homepage
• 代码/code：None

Self-Calibrating Neural Radiance Fields

• 论文/paper：https://arxiv.org/abs/2108.13826
• 代码/code：https://github.com/POSTECH-CVLab/SCNeRF

UNISURF: Unifying Neural Implicit Surfaces and Radiance Fields for Multi-View Reconstruction (Oral)

• 论文/paper：https://arxiv.org/abs/2104.10078 | 主页/Homepage

• 代码/code：None

返回目录/back

Shadow Removal

CANet: A Context-Aware Network for Shadow Removal

• 论文/paper：https://arxiv.org/abs/2108.09894v1
• 代码/code：None

返回目录/back

Image Retrieval

DOLG: Single-Stage Image Retrieval with Deep Orthogonal Fusion of Local and Global Features

• 论文/paper：https://arxiv.org/abs/2108.02927
• 代码/code：None

Image Retrieval on Real-life Images with Pre-trained Vision-and-Language Models

• 论文/paper：https://arxiv.org/abs/2108.04024
• 代码/code：https://github.com/Cuberick-Orion/CIRR

Self-supervised Product Quantization for Deep Unsupervised Image Retrieval

• 论文/paper：https://arxiv.org/abs/2109.02244
• 代码/code：None

返回目录/back

超分辨/Super-Resolution

Designing a Practical Degradation Model for Deep Blind Image Super-Resolution

• 论文/paper：https://arxiv.org/pdf/2103.14006.pdf
• 代码/code：https://github.com/cszn/BSRGAN

Dual-Camera Super-Resolution with Aligned Attention Modules

• 论文/paper：https://arxiv.org/abs/2109.01349

• 代码/code：None

Generalized Real-World Super-Resolution through Adversarial Robustness

• 论文/paper：https://arxiv.org/abs/2108.11505

• 代码/code：None

Learning for Scale-Arbitrary Super-Resolution from Scale-Specific Networks

• 论文/paper：https://arxiv.org/abs/2004.03791

• 代码/code：https://github.com/LongguangWang/ArbSR

Overfitting the Data: Compact Neural Video Delivery via Content-aware Feature Modulation

• 论文/paper：None

• 代码/code： https://github.com/Anonymous-iccv2021-paper3163/CaFM-Pytorch

返回目录/back

Image Reconstruction

Equivariant Imaging: Learning Beyond the Range Space (Oral)

• 论文/paper：https://arxiv.org/abs/2103.14756
• 代码/code：https://github.com/edongdongchen/EI

Spatially-Adaptive Image Restoration using Distortion-Guided Networks

• 论文/paper：https://arxiv.org/abs/2108.08617
• 代码/code：https://github.com/human-analysis/spatially-adaptive-image-restoration

返回目录/back

Image Deblurring

Defocus Map Estimation and Deblurring from a Single Dual-Pixel Image (Oral)

• 论文/paper：https://arxiv.org/abs/2110.05655
• 代码/code：None

SDWNet: A Straight Dilated Network with Wavelet Transformation for Image Deblurring

• 论文/paper：https://arxiv.org/abs/2110.05803
• 代码/code：https://github.com/FlyEgle/SDWNet

Single Image Defocus Deblurring Using Kernel-Sharing Parallel Atrous Convolutions

• 论文/paper：https://arxiv.org/abs/2108.09108
• 代码/code：None

返回目录/back

Image Denoising

Deep Reparametrization of Multi-Frame Super-Resolution and Denoising (Oral)

• 论文/paper：https://arxiv.org/abs/2108.08286
• 代码/code：None

Eformer: Edge Enhancement based Transformer for Medical Image Denoising

• 论文/paper：https://arxiv.org/abs/2109.08044
• 代码/code：None

**ILVR: Conditioning Method for Denoising Diffusion Probabilistic Models **Oral

• 论文/paper：https://arxiv.org/abs/2108.02938
• 代码/code：None

Rethinking Deep Image Prior for Denoising

• 论文/paper：https://arxiv.org/abs/2108.12841
• 代码/code：None

Rethinking Noise Synthesis and Modeling in Raw Denoising

• 文/paper：https://arxiv.org/abs/2110.04756
• 代码/code：None

返回目录/back

Image Desnowing

ALL Snow Removed: Single Image Desnowing Algorithm Using Hierarchical Dual-tree Complex Wavelet Representation and Contradict Channel Loss

• 论文/paper：None
• 代码/code：https://github.com/weitingchen83/ICCV2021-Single-Image-Desnowing-HDCWNet

返回目录/back

Image Enhancement

Gap-closing Matters: Perceptual Quality Assessment and Optimization of Low-Light Image Enhancement

• 论文/paper：None
• 代码/code：https://github.com/Baoliang93/Gap-closing-Matters

Real-time Image Enhancer via Learnable Spatial-aware 3D Lookup Tables

• 论文/paper：https://arxiv.org/abs/2108.08697
• 代码/code：None

返回目录/back

Image Matching

Effect of Parameter Optimization on Classical and Learning-based Image Matching Methods

• 论文/paper：https://arxiv.org/abs/2108.08179
• 代码/code：None

Viewpoint Invariant Dense Matching for Visual Geolocalization

• 论文/paper：https://arxiv.org/abs/2109.09827
• 代码/code：https://github.com/gmberton/geo_warp

返回目录/back

Image Quality

MUSIQ: Multi-scale Image Quality Transformer

• 论文/paper：https://arxiv.org/abs/2108.05997
• 代码/code：None

返回目录/back

Image Compression

Dense Deep Unfolding Network with 3D-CNN Prior for Snapshot Compressive Imaging

• 论文/paper：https://arxiv.org/abs/2109.06548
• 代码/code：https://github.com/jianzhangcs/SCI3D

Variable-Rate Deep Image Compression through Spatially-Adaptive Feature Transform

• 论文/paper：https://arxiv.org/abs/2108.09551v1
• 代码/code：https://github.com/micmic123/QmapCompression

返回目录/back

Image Restoration

Dynamic Attentive Graph Learning for Image Restoration

• 论文/paper：https://arxiv.org/abs/2109.06620
• 代码/code：https://github.com/jianzhangcs/DAGL

Towards Flexible Blind JPEG Artifacts Removal

• 论文/paper：https://arxiv.org/abs/2109.14573
• 代码/code：https://github.com/jiaxi-jiang/FBCNN

返回目录/back

Image Inpainting

Image Inpainting via Conditional Texture and Structure Dual Generation

• 论文/paper：https://arxiv.org/abs/2108.09760v1
• 代码/code：https://github.com/Xiefan-Guo/CTSDG

返回目录/back

Video Inpainting

FuseFormer: Fusing Fine-Grained Information in Transformers for Video Inpainting

• 论文/paper：https://arxiv.org/abs/2108.01912
• 代码/code：None

Internal Video Inpainting by Implicit Long-range Propagation

• 论文/paper：https://arxiv.org/abs/2108.01912
• 代码/code：None

Occlusion-Aware Video Object Inpainting

• 论文/paper：https://arxiv.org/abs/2108.06765
• 代码/code：None

返回目录/back

Video Recognition

Searching for Two-Stream Models in Multivariate Space for Video Recognition

• 论文/paper：https://arxiv.org/abs/2108.12957
• 代码/code：None

返回目录/back

Visual Question Answering

Weakly Supervised Relative Spatial Reasoning for Visual Question Answering

• 论文/paper：https://arxiv.org/abs/2109.01934
• 代码/code：https://github.com/pratyay-banerjee/weak_sup_vqa

返回目录/back

Matching

Multi-scale Matching Networks for Semantic Correspondence

• 论文/paper：https://arxiv.org/abs/2108.00211
• 代码/code：None

返回目录/back

人机交互/Hand-object Interaction

✔️CPF: Learning a Contact Potential Field to Model the Hand-object Interaction

• 论文/paper：https://arxiv.org/abs/2012.00924
• 代码/code：https://github.com/lixiny/CPF

Exploiting Scene Graphs for Human-Object Interaction Detection

• 论文/paper：https://arxiv.org/abs/2108.08584
• 代码/code：https://github.com/ht014/SG2HOI

Spatially Conditioned Graphs for Detecting Human–Object Interactions

• 论文/paper：https://arxiv.org/pdf/2012.06060.pdf
• 代码/code：https://github.com/fredzzhang/spatially-conditioned-graphs

Virtual Multi-Modality Self-Supervised Foreground Matting for Human-Object Interaction

• 论文/paper：https://arxiv.org/abs/2110.03278
• 代码/code：None

返回目录/back

视线估计/Gaze Estimation

Generalizing Gaze Estimation with Outlier-guided Collaborative Adaptation

• 论文/paper：https://arxiv.org/abs/2107.13780 | 主页/Homepage

• 代码/code：https://github.com/DreamtaleCore/PnP-GA

返回目录/back

Contrastive-Learning

Attentive and Contrastive Learning for Joint Depth and Motion Field Estimation

• 论文/paper：https://arxiv.org/abs/2110.06853

• 代码/code：None

Improving Contrastive Learning by Visualizing Feature Transformation

• 论文/paper：https://arxiv.org/abs/2108.02982

• 代码/code：https://github.com/DTennant/CL-Visualizing-Feature-Transformation

Social NCE: Contrastive Learning of Socially-aware Motion Representations

• 论文/paper：https://arxiv.org/abs/2012.11717

• 代码/code：https://github.com/vita-epfl/social-nce-crowdnav

Parametric Contrastive Learning

• 论文/paper：https://arxiv.org/abs/2107.12028

• 代码/code：https://github.com/jiequancui/Parametric-Contrastive-Learning

返回目录/back

Graph Convolution Networks

MSR-GCN: Multi-Scale Residual Graph Convolution Networks for Human Motion Prediction

• 论文/paper：None
• 代码/code：https://github.com/Droliven/MSRGCN

返回目录/back

模型压缩/Compress

GDP: Stabilized Neural Network Pruning via Gates with Differentiable Polarization

• 论文/paper：https://arxiv.org/abs/2109.02220

• 代码/code：None

Sub-bit Neural Networks: Learning to Compress and Accelerate Binary Neural Networks

• 论文/paper：https://arxiv.org/abs/2110.09195

• 代码/code：https://github.com/yikaiw/SNN

返回目录/back

Quantization

Cluster-Promoting Quantization with Bit-Drop for Minimizing Network Quantization Loss

• 论文/paper：https://arxiv.org/abs/2109.02100
• 代码/code：None

Distance-aware Quantization

• 论文/paper：https://arxiv.org/abs/2108.06983
• 代码/code：None

Dynamic Network Quantization for Efficient Video Inference

• 论文/paper：https://arxiv.org/abs/2108.10394
• 代码/code：None

Generalizable Mixed-Precision Quantization via Attribution Rank Preservation

• 论文/paper：https://arxiv.org/abs/2108.02720
• 代码/code：https://github.com/ZiweiWangTHU/GMPQ

Towards Mixed-Precision Quantization of Neural Networks via Constrained Optimization

• 论文/paper：https://arxiv.org/abs/2110.06554
• 代码/code：None

返回目录/back

Knowledge Distillation

Deep Structured Instance Graph for Distilling Object Detectors

• 论文/paper：https://arxiv.org/abs/2109.12862

• 代码/code：https://github.com/dvlab-research/Dsig

Distilling Holistic Knowledge with Graph Neural Networks

• 论文/paper：https://arxiv.org/abs/2108.05507
• 代码/code：https://github.com/wyc-ruiker/HKD

Lipschitz Continuity Guided Knowledge Distillation

• 论文/paper：https://arxiv.org/abs/2108.12905

• 代码/code：None

G-DetKD: Towards General Distillation Framework for Object Detectors via Contrastive and Semantic-guided Feature Imitation

• 论文/paper：https://arxiv.org/abs/2108.07482
• 代码/code：None

Self Supervision to Distillation for Long-Tailed Visual Recognition

• 论文/paper：https://arxiv.org/abs/2109.04075
• 代码/code：https://github.com/MCG-NJU/SSD-LT

返回目录/back

点云/Point Cloud

A Robust Loss for Point Cloud Registration

• 论文/paper：https://arxiv.org/abs/2108.11682

• 代码/code：None

A Technical Survey and Evaluation of Traditional Point Cloud Clustering Methods for LiDAR Panoptic Segmentation

• 论文/paper：https://arxiv.org/abs/2108.09522v1

• 代码/code：None

(Just) A Spoonful of Refinements Helps the Registration Error Go Down Oral

• 论文/paper：https://arxiv.org/abs/2108.03257

• 代码/code：None

ABD-Net: Attention Based Decomposition Network for 3D Point Cloud Decomposition

• 论文/paper：https://arxiv.org/abs/2108.04221
• 代码/code：None

AdaFit: Rethinking Learning-based Normal Estimation on Point Clouds

• 论文/paper：https://arxiv.org/abs/2108.05836
• 代码/code：None

Box-Aware Feature Enhancement for Single Object Tracking on Point Clouds

• 论文/paper：https://arxiv.org/abs/2108.04728
• 代码/code：None

CPFN: Cascaded Primitive Fitting Networks for High-Resolution Point Clouds

• 论文/paper：https://arxiv.org/abs/2109.00113
• 代码/code：None

Deep Models with Fusion Strategies for MVP Point Cloud Registration

• 论文/paper：https://arxiv.org/abs/2110.09129
• 代码/code：None

DRINet: A Dual-Representation Iterative Learning Network for Point Cloud Segmentation

• 论文/paper：https://arxiv.org/abs/2108.04023
• 代码/code：None

Guided Point Contrastive Learning for Semi-supervised Point Cloud Semantic Segmentation

• 论文/paper：https://arxiv.org/abs/2110.08188
• 代码/code：None

Learning Inner-Group Relations on Point Clouds

• 论文/paper：https://arxiv.org/abs/2108.12468
• 代码/code：None

InstanceRefer: Cooperative Holistic Understanding for Visual Grounding on Point Clouds through Instance Multi-level Contextual Referring

• 论文/paper：https://arxiv.org/pdf/2103.01128.pdf
• 代码/code：https://github.com/CurryYuan/InstanceRefer

ME-PCN: Point Completion Conditioned on Mask Emptiness

• 论文/paper：https://arxiv.org/abs/2108.08187
• 代码/code：None

MVP Benchmark: Multi-View Partial Point Clouds for Completion and Registration

• 论文/paper：None |主页/Homepage
• 代码/code：https://github.com/paul007pl/MVP_Benchmark

Out-of-Core Surface Reconstruction via Global TGV Minimization

• 论文/paper：https://arxiv.org/abs/2107.14790
• 代码/code：None

PCAM: Product of Cross-Attention Matrices for Rigid Registration of Point Clouds

• 论文/paper：https://arxiv.org/abs/2110.01269
• 代码/code：https://github.com/valeoai/PCAM

PICCOLO: Point Cloud-Centric Omnidirectional Localization

• 论文/paper：https://arxiv.org/abs/2108.06545
• 代码/code：None

Point Cloud Augmentation with Weighted Local Transformations

• 论文/paper：https://arxiv.org/abs/2110.05379
• 代码/code：None

PoinTr: Diverse Point Cloud Completion with Geometry-Aware Transformers (Oral)

• 论文/paper：https://arxiv.org/abs/2108.08839
• 代码/code：https://github.com/yuxumin/PoinTr

ReDAL: Region-based and Diversity-aware Active Learning for Point Cloud Semantic Segmentation

• 论文/paper：https://arxiv.org/abs/2107.11769
• 代码/code：None

Sampling Network Guided Cross-Entropy Method for Unsupervised Point Cloud Registration

• 论文/paper：https://arxiv.org/abs/2109.06619
• 代码/code：None

SnowflakeNet: Point Cloud Completion by Snowflake Point Deconvolution with Skip-Transformer

• 论文/paper：https://arxiv.org/abs/2108.04444
• 代码/code：https://github.com/AllenXiangX/SnowflakeNet

Spatio-temporal Self-Supervised Representation Learning for 3D Point Clouds

• 论文/paper：https://arxiv.org/abs/2109.00179
• 代码/code：None

Towards Efficient Point Cloud Graph Neural Networks Through Architectural Simplification

• 论文/paper：https://arxiv.org/abs/2108.06317
• 代码/code：None

Unsupervised Learning of Fine Structure Generation for 3D Point Clouds by 2D Projection Matching

• 论文/paper：https://arxiv.org/abs/2108.03746
• 代码/code：https://github.com/chenchao15/2D

Unsupervised Point Cloud Pre-Training via View-Point Occlusion, Completion

• 论文/paper：https://arxiv.org/abs/2010.01089 |主页/Homepage
• 代码/code：https://github.com/hansen7/OcCo

Vis2Mesh: Efficient Mesh Reconstruction from Unstructured Point Clouds of Large Scenes with Learned Virtual View Visibility

• 论文/paper：https://arxiv.org/abs/2108.08378
• 代码/code：https://github.com/GDAOSU/vis2mesh

Voxel-based Network for Shape Completion by Leveraging Edge Generation

• 论文/paper：https://arxiv.org/abs/2108.09936v1
• 代码/code：https://github.com/xiaogangw/VE-PCN

Walk in the Cloud: Learning Curves for Point Clouds Shape Analysis

• 论文/paper：https://arxiv.org/abs/2105.01288v1| 主页/Homepage

• 代码/code：https://github.com/tiangexiang/CurveNet

返回目录/back

3D reconstruction

3D Shapes Local Geometry Codes Learning with SDF

• 论文/paper：https://arxiv.org/abs/2108.08593
• 代码/code：None

3DIAS: 3D Shape Reconstruction with Implicit Algebraic Surfaces

• 论文/paper：https://arxiv.org/abs/2108.08653
• 代码/code：https://myavartanoo.github.io/3dias/

DensePose 3D: Lifting Canonical Surface Maps of Articulated Objects to the Third Dimension

• 论文/paper：https://arxiv.org/abs/2109.00033
• 代码/code：None

Learning Anchored Unsigned Distance Functions with Gradient Direction Alignment for Single-view Garment Reconstruction

• 论文/paper：https://arxiv.org/abs/2108.08478
• 代码/code：None

Pixel-Perfect Structure-from-Motion with Featuremetric Refinement(Oral)

• 论文/paper：https://arxiv.org/abs/2108.08291
• 代码/code：https://github.com/cvg/pixel-perfect-sfm

VolumeFusion: Deep Depth Fusion for 3D Scene Reconstruction

• 论文/paper：https://arxiv.org/abs/2108.08623
• 代码/code：None

返回目录/back

字体生成/Font Generation

✔️Multiple Heads are Better than One: Few-shot Font Generation with Multiple Localized Experts

• 论文/paper：https://arxiv.org/abs/2104.00887

• 代码/code：https://github.com/clovaai/mxfont

返回目录/back

文本检测 / Text Detection

Adaptive Boundary Proposal Network for Arbitrary Shape Text Detection

• 论文/paper：https://arxiv.org/abs/2107.12664
• 代码/code：https://github.com/GXYM/TextBPN

返回目录/back

文本识别 / Text Recognition

From Two to One: A New Scene Text Recognizer with Visual Language Modeling Network

• 论文/paper：https://arxiv.org/abs/2108.09661v1
• 代码/code：https://github.com/wangyuxin87/VisionLAN

Joint Visual Semantic Reasoning: Multi-Stage Decoder for Text Recognition

• 论文/paper：https://arxiv.org/abs/2107.12090
• 代码/code：None

返回目录/back

Scene Text Recognizer

Data Augmentation for Scene Text Recognition

• 论文/paper：https://arxiv.org/abs/2108.06949

• 代码/code：https://github.com/roatienza/straug

From Two to One: A New Scene Text Recognizer with Visual Language Modeling Network

• 论文/paper：None

• 代码/code：https://github.com/wangyuxin87/VisionLAN

返回目录/back

Autonomous-Driving

End-to-End Urban Driving by Imitating a Reinforcement Learning Coach

• 论文/paper：https://arxiv.org/abs/2108.08265
• 代码/code：None

FOVEA: Foveated Image Magnification for Autonomous Navigation

• 论文/paper：https://arxiv.org/abs/2108.12102v1
• 代码/code：https://www.cs.cmu.edu/~mengtial/proj/fovea/

Learning to drive from a world on rails

• 论文/paper：https://arxiv.org/abs/2105.00636
• 代码/code：https://arxiv.org/abs/2105.00636

MAAD: A Model and Dataset for "Attended Awareness" in Driving

• 论文/paper：https://arxiv.org/abs/2110.08610
• 代码/code：https://github.com/ToyotaResearchInstitute/att-aware/

MultiSiam: Self-supervised Multi-instance Siamese Representation Learning for Autonomous Driving

• 论文/paper：https://arxiv.org/abs/2108.12178v1
• 代码/code：https://github.com/KaiChen1998/MultiSiam

NEAT: Neural Attention Fields for End-to-End Autonomous Driving

• 论文/paper：https://arxiv.org/abs/2109.04456
• 代码/code：None

Road-Challenge-Event-Detection-for-Situation-Awareness-in-Autonomous-Driving

• 论文/paper：None
• 代码/code：https://github.com/Trevorchenmsu/Road-Challenge-Event-Detection-for-Situation-Awareness-in-Autonomous-Driving

Safety-aware Motion Prediction with Unseen Vehicles for Autonomous Driving

• 论文/paper：https://arxiv.org/abs/2109.01510

• 代码/code：https://github.com/xrenaa/Safety-Aware-Motion-Prediction

返回目录/back

Visdrone_detection

ICCV2021_Visdrone_detection

• 论文/paper：None

• 代码/code：https://github.com/Gumpest/ICCV2021_Visdrone_detection

返回目录/back

Anomaly Detection

DRÆM -- A discriminatively trained reconstruction embedding for surface anomaly detection

• 论文/paper：https://arxiv.org/abs/2108.07610

• 代码/code：None

Weakly-supervised Video Anomaly Detection with Robust Temporal Feature Magnitude Learning

• 论文/paper：https://arxiv.org/pdf/2101.10030.pdf

• 代码/code：https://github.com/tianyu0207/RTFM

其他/Others

Cross-Camera Convolutional Color Constancy

• 论文/paper：https://arxiv.org/abs/2011.11164

• 代码/code：https://github.com/mahmoudnafifi/C5

Learnable Boundary Guided Adversarial Training

• 论文/paper：https://arxiv.org/abs/2011.11164

• 代码/code：https://github.com/FPNAS/LBGAT

Prior-Enhanced network with Meta-Prototypes (PEMP)

• 论文/paper：None
• 代码/code：https://github.com/PaperSubmitAAAA/ICCV2021-2337

MDETR -- Modulated Detection for End-to-End Multi-Modal Understanding

• 论文/paper：https://arxiv.org/abs/2104.12763 | 主页/Homepage
• 代码/code：https://github.com/ashkamath/mdetr

Generalized-Shuffled-Linear-Regression （Oral）

• 论文/paper：https://drive.google.com/file/d/1Qu21VK5qhCW8WVjiRnnBjehrYVmQrDNh/view
• 代码/code：https://github.com/SILI1994/Generalized-Shuffled-Linear-Regression

VLGrammar: Grounded Grammar Induction of Vision and Language

• 论文/paper：https://arxiv.org/abs/2103.12975
• 代码/code：https://github.com/evelinehong/VLGrammar

A New Journey from SDRTV to HDRTV

• 论文/paper：None
• 代码/code：https://github.com/chxy95/HDRTVNet

IICNet: A Generic Framework for Reversible Image Conversion

• 论文/paper：None
• 代码/code：https://github.com/felixcheng97/IICNet

Structure-Preserving Deraining with Residue Channel Prior Guidance

• 论文/paper：None
• 代码/code：https://github.com/Joyies/SPDNet

Learning with Noisy Labels via Sparse Regularization

• 论文/paper：https://arxiv.org/abs/2108.00192
• 代码/code：https://github.com/hitcszx/lnl_sr

Neural Strokes: Stylized Line Drawing of 3D Shapes

• 论文/paper：None
• 代码/code：https://github.com/DifanLiu/NeuralStrokes

COOKIE: Contrastive Cross-Modal Knowledge Sharing Pre-training for Vision-Language Representation

• 论文/paper：None
• 代码/code：https://github.com/kywen1119/COOKIE

RINDNet: Edge Detection for Discontinuity in Reflectance, Illumination, Normal and Depth

• 论文/paper：https://arxiv.org/abs/2108.00616
• 代码/code：None

ELLIPSDF: Joint Object Pose and Shape Optimization with a Bi-level Ellipsoid and Signed Distance Function Description

• 论文/paper：https://arxiv.org/abs/2108.00355
• 代码/code：None

Unlimited Neighborhood Interaction for Heterogeneous Trajectory Prediction

• 论文/paper：https://arxiv.org/abs/2108.00238
• 代码/code：None

CanvasVAE: Learning to Generate Vector Graphic Documents

• 论文/paper：https://arxiv.org/abs/2108.01249
• 代码/code：None

Refining activation downsampling with SoftPool

• 论文/paper：https://arxiv.org/abs/2101.00440
• 代码/code：https://github.com/alexandrosstergiou/SoftPool

Aligning Latent and Image Spaces to Connect the Unconnectable

• 论文/paper：https://arxiv.org/abs/2104.06954 | 主页/Homepage
• 代码/code：https://github.com/universome/alis

Unifying Nonlocal Blocks for Neural Networks

• 论文/paper：https://arxiv.org/abs/2108.02451
• 代码/code：None

SLAMP: Stochastic Latent Appearance and Motion Prediction

• 论文/paper：https://arxiv.org/abs/2108.02760
• 代码/code：None

TransForensics: Image Forgery Localization with Dense Self-Attention

• 论文/paper：https://arxiv.org/abs/2108.03871
• 代码/code：None

Learning Facial Representations from the Cycle-consistency of Face

• 论文/paper：https://arxiv.org/abs/2108.03427
• 代码/code：https://github.com/JiaRenChang/FaceCycle

NASOA: Towards Faster Task-oriented Online Fine-tuning with a Zoo of Models

• 论文/paper：https://arxiv.org/abs/2108.03434
• 代码/code：None

Impact of Aliasing on Generalization in Deep Convolutional Networks

• 论文/paper：https://arxiv.org/abs/2108.03489
• 代码/code：None

Learning Canonical 3D Object Representation for Fine-Grained Recognition

• 论文/paper：https://arxiv.org/abs/2108.04628
• 代码/code：None

UniNet: A Unified Scene Understanding Network and Exploring Multi-Task Relationships through the Lens of Adversarial Attacks

• 论文/paper：https://arxiv.org/abs/2108.04584
• 代码/code：None

SUNet: Symmetric Undistortion Network for Rolling Shutter Correction

• 论文/paper：https://arxiv.org/abs/2108.04775
• 代码/code：None

Learning to Cut by Watching Movies

• 论文/paper：https://arxiv.org/abs/2108.04294
• 代码/code：https://github.com/PardoAlejo/LearningToCut

Continual Neural Mapping: Learning An Implicit Scene Representation from Sequential Observations

• 论文/paper：https://arxiv.org/abs/2108.05851
• 代码/code：None

Towers of Babel: Combining Images, Language, and 3D Geometry for Learning Multimodal Vision

• 论文/paper：https://arxiv.org/abs/2108.05863 |主页/Homepage
• 代码/code：https://github.com/tgxs002/wikiscenes

Towards Interpretable Deep Metric Learning with Structural Matching

• 论文/paper：https://arxiv.org/abs/2108.05889
• 代码/code：https://github.com/wl-zhao/DIML

m-RevNet: Deep Reversible Neural Networks with Momentum

• 论文/paper：https://arxiv.org/abs/2108.05862
• 代码/code：None

DiagViB-6: A Diagnostic Benchmark Suite for Vision Models in the Presence of Shortcut and Generalization Opportunities

• 论文/paper：https://arxiv.org/abs/2108.05779
• 代码/code：None

perf4sight: A toolflow to model CNN training performance on Edge GPUs

• 论文/paper：https://arxiv.org/abs/2108.05580
• 代码/code：None

MT-ORL: Multi-Task Occlusion Relationship Learning

• 论文/paper：https://arxiv.org/abs/2108.05722
• 代码/code：https://github.com/fengpanhe/MT-ORL

ProAI: An Efficient Embedded AI Hardware for Automotive Applications - a Benchmark Study

• 论文/paper：https://arxiv.org/abs/2108.05170
• 代码/code：None

SPACE: A Simulator for Physical Interactions and Causal Learning in 3D Environments

• 论文/paper：https://arxiv.org/abs/2108.06180
• 代码/code：https://github.com/jiafei1224/SPACE

CODEs: Chamfer Out-of-Distribution Examples against Overconfidence Issue

• 论文/paper：https://arxiv.org/abs/2108.06024
• 代码/code：None

Towards Real-World Prohibited Item Detection: A Large-Scale X-ray Benchmark

• 论文/paper：https://arxiv.org/abs/2108.07020
• 代码/code：None

Pixel Difference Networks for Efficient Edge Detection

• 论文/paper：https://arxiv.org/abs/2108.07009
• 代码/code：https://github.com/zhuoinoulu/pidinet

Online Continual Learning For Visual Food Classification

• 论文/paper：https://arxiv.org/abs/2108.06781
• 代码/code：None

DICOM Imaging Router: An Open Deep Learning Framework for Classification of Body Parts from DICOM X-ray Scans

• 论文/paper：https://arxiv.org/abs/2108.06490 |主页/Homepage
• 代码/code：None

PIT: Position-Invariant Transform for Cross-FoV Domain Adaptation

• 论文/paper：https://arxiv.org/abs/2108.07142
• 代码/code：https://github.com/sheepooo/PIT-Position-Invariant-Transform

Learning to Automatically Diagnose Multiple Diseases in Pediatric Chest Radiographs Using Deep Convolutional Neural Networks

• 论文/paper：https://arxiv.org/abs/2108.06486
• 代码/code：None

FaPN: Feature-aligned Pyramid Network for Dense Image Prediction

• 论文/paper：https://arxiv.org/abs/2108.07058
• 代码/code：https://github.com/EMI-Group/FaPN

Finding Representative Interpretations on Convolutional Neural Networks

• 论文/paper：https://arxiv.org/abs/2108.06384
• 代码/code：None

Investigating transformers in the decomposition of polygonal shapes as point collections

• 论文/paper：https://arxiv.org/abs/2108.07533
• 代码/code：None

Self-Supervised Pretraining and Controlled Augmentation Improve Rare Wildlife Recognition in UAV Images

• 论文/paper：https://arxiv.org/abs/2108.07582
• 代码/code：None

Group-aware Contrastive Regression for Action Quality Assessment

• 论文/paper：https://arxiv.org/abs/2108.07797
• 代码/code：None

End-to-End Dense Video Captioning with Parallel Decoding

• 论文/paper：https://arxiv.org/abs/2108.07781
• 代码/code：https://github.com/ttengwang/PDVC

PR-RRN: Pairwise-Regularized Residual-Recursive Networks for Non-rigid Structure-from-Motion

• 论文/paper：https://arxiv.org/abs/2108.07506
• 代码/code：None

Scene Designer: a Unified Model for Scene Search and Synthesis from Sketch

• 论文/paper：https://arxiv.org/abs/2108.07353
• 代码/code：None

Structured Outdoor Architecture Reconstruction by Exploration and Classification

• 论文/paper：https://arxiv.org/abs/2108.07990
• 代码/code：None

Learning RAW-to-sRGB Mappings with Inaccurately Aligned Supervision

• 论文/paper：https://arxiv.org/abs/2108.08119
• 代码/code：https://github.com/cszhilu1998/RAW-to-sRGB

Overfitting the Data: Compact Neural Video Delivery via Content-aware Feature Modulation

• 论文/paper：https://arxiv.org/abs/2108.08202
• 代码/code：https://github.com/Neural-video-delivery/CaFM-Pytorch-ICCV2021

Deep Hybrid Self-Prior for Full 3D Mesh Generation

• 论文/paper：https://arxiv.org/abs/2108.08017
• 代码/code：None

FACIAL: Synthesizing Dynamic Talking Face with Implicit Attribute Learning

• 论文/paper：https://arxiv.org/abs/2108.07938
• 代码/code：None

Thermal Image Processing via Physics-Inspired Deep Networks

• 论文/paper：https://arxiv.org/abs/2108.07973
• 代码/code：None

A New Journey from SDRTV to HDRTV

• 论文/paper：https://arxiv.org/abs/2108.07978
• 代码/code：https://github.com/chxy95/HDRTVNet

Global Pooling, More than Meets the Eye: Position Information is Encoded Channel-Wise in CNNs

• 论文/paper：https://arxiv.org/abs/2108.07884
• 代码/code：None

Speech Drives Templates: Co-Speech Gesture Synthesis with Learned Templates

• 论文/paper：https://arxiv.org/abs/2108.08020
• 代码/code：None

LOKI: Long Term and Key Intentions for Trajectory Prediction

• 论文/paper：https://arxiv.org/abs/2108.08236
• 代码/code：None

Stochastic Scene-Aware Motion Prediction

• 论文/paper：https://arxiv.org/abs/2108.08284
• 代码/code：https://samp.is.tue.mpg.de/

Exploiting Multi-Object Relationships for Detecting Adversarial Attacks in Complex Scenes

• 论文/paper：https://arxiv.org/abs/2108.08421
• 代码/code：None

Social Fabric: Tubelet Compositions for Video Relation Detection

• 论文/paper：https://arxiv.org/abs/2108.08363
• 代码/code：https://github.com/shanshuo/Social-Fabric

Causal Attention for Unbiased Visual Recognition

• 论文/paper：https://arxiv.org/abs/2108.08782
• 代码/code：https://github.com/Wangt-CN/CaaM

Universal Cross-Domain Retrieval: Generalizing Across Classes and Domains

• 论文/paper：https://arxiv.org/abs/2108.08356
• 代码/code：None

Amplitude-Phase Recombination: Rethinking Robustness of Convolutional Neural Networks in Frequency Domain

• 论文/paper：https://arxiv.org/abs/2108.08487
• 代码/code：None

Learning to Match Features with Seeded Graph Matching Network

• 论文/paper：https://arxiv.org/abs/2108.08771
• 代码/code：https://github.com/vdvchen/SGMNet

A Unified Objective for Novel Class Discovery

• 论文/paper：https://arxiv.org/abs/2108.08536
• 代码/code：https://github.com/DonkeyShot21/UNO

How to cheat with metrics in single-image HDR reconstruction

• 论文/paper：https://arxiv.org/abs/2108.08713
• 代码/code：None

Towards Understanding the Generative Capability of Adversarially Robust Classifiers （Oral）

• 论文/paper：https://arxiv.org/abs/2108.09093
• 代码/code：None

Airbert: In-domain Pretraining for Vision-and-Language Navigation

• 论文/paper：https://arxiv.org/abs/2108.09105
• 代码/code：None

Out-of-boundary View Synthesis Towards Full-Frame Video Stabilization

• 论文/paper：https://arxiv.org/abs/2108.09041
• 代码/code：https://github.com/Annbless/OVS_Stabilization

PatchMatch-RL: Deep MVS with Pixelwise Depth, Normal, and Visibility

• 论文/paper：https://arxiv.org/abs/2108.08943
• 代码/code：None

Continual Learning for Image-Based Camera Localization

• 论文/paper：https://arxiv.org/abs/2108.09112
• 代码/code：None

Online Continual Learning with Natural Distribution Shifts: An Empirical Study with Visual Data

• 论文/paper：https://arxiv.org/abs/2108.09020
• 代码/code：https://github.com/IntelLabs/continuallearning

Detecting and Segmenting Adversarial Graphics Patterns from Images

• 论文/paper：https://arxiv.org/abs/2108.09383v1
• 代码/code：None

TACo: Token-aware Cascade Contrastive Learning for Video-Text Alignment

• 论文/paper：https://arxiv.org/abs/2108.09980v1
• 代码/code：None

BlockCopy: High-Resolution Video Processing with Block-Sparse Feature Propagation and Online Policies

• 论文/paper：https://arxiv.org/abs/2108.09376v1
• 代码/code：None

Learning Signed Distance Field for Multi-view Surface Reconstruction (Oral)

• 论文/paper：https://arxiv.org/abs/2108.09964v1
• 代码/code：None

Deep Relational Metric Learning

• 论文/paper：https://arxiv.org/abs/2108.10026v1
• 代码/code：https://github.com/zbr17/DRML

Ranking Models in Unlabeled New Environments

• 论文/paper：https://arxiv.org/abs/2108.10310v1
• 代码/code：https://github.com/sxzrt/Proxy-Set

Patch2CAD: Patchwise Embedding Learning for In-the-Wild Shape Retrieval from a Single Image

• 论文/paper：https://arxiv.org/abs/2108.09368v1
• 代码/code：None

LSD-StructureNet: Modeling Levels of Structural Detail in 3D Part Hierarchies

• 论文/paper：https://arxiv.org/abs/2108.13459
• 代码/code：None

BiaSwap: Removing dataset bias with bias-tailored swapping augmentation

• 论文/paper：https://arxiv.org/abs/2108.10008v1
• 代码/code：None

LoOp: Looking for Optimal Hard Negative Embeddings for Deep Metric Learning

• 论文/paper：https://arxiv.org/abs/2108.09335v1
• 代码/code：None

Learning of Visual Relations: The Devil is in the Tails

• 论文/paper：https://arxiv.org/abs/2108.09668v1
• 代码/code：None

Bridging Unsupervised and Supervised Depth from Focus via All-in-Focus Supervision

• 论文/paper：https://arxiv.org/abs/2108.10843
• 代码/code：https://github.com/albert100121/AiFDepthNet

Support-Set Based Cross-Supervision for Video Grounding

• 论文/paper：https://arxiv.org/abs/2108.10576
• 代码/code：None

Fast Robust Tensor Principal Component Analysis via Fiber CUR Decomposition

• 论文/paper：https://arxiv.org/abs/2108.10448
• 代码/code：None

Improving Generalization of Batch Whitening by Convolutional Unit Optimization

• 论文/paper：https://arxiv.org/abs/2108.10629
• 代码/code：None

CSG-Stump: A Learning Friendly CSG-Like Representation for Interpretable Shape Parsing

• 论文/paper：https://arxiv.org/abs/2108.11305 |主页/Homepage
• 代码/code：https://github.com/kimren227/CSGStumpNet

NGC: A Unified Framework for Learning with Open-World Noisy Data

• 论文/paper：https://arxiv.org/abs/2108.11035
• 代码/code：None

LocTex: Learning Data-Efficient Visual Representations from Localized Textual Supervision

• 论文/paper：https://arxiv.org/abs/2108.11950
• 代码/code：https://loctex.mit.edu/

The Surprising Effectiveness of Visual Odometry Techniques for Embodied PointGoal Navigation

• 论文/paper：https://arxiv.org/abs/2108.11550
• 代码/code：None

Learning Cross-modal Contrastive Features for Video Domain Adaptation

• 论文/paper：https://arxiv.org/abs/2108.11974v1
• 代码/code：None

Lifelong Infinite Mixture Model Based on Knowledge-Driven Dirichlet Process

• 论文/paper：https://arxiv.org/abs/2108.12278v1
• 代码/code：https://github.com/dtuzi123/Lifelong-infinite-mixture-model

A Dual Adversarial Calibration Framework for Automatic Fetal Brain Biometry

• 论文/paper：https://arxiv.org/abs/2108.12719
• 代码/code：None

LUAI Challenge 2021 on Learning to Understand Aerial Images

• 论文/paper：https://arxiv.org/abs/2108.13246
• 代码/code：None

Embedding Novel Views in a Single JPEG Image

• 论文/paper：https://arxiv.org/abs/2108.13003
• 代码/code：None

Learning to Discover Reflection Symmetry via Polar Matching Convolution

• 论文/paper：https://arxiv.org/abs/2108.12952
• 代码/code：None

Deep 3D Mask Volume for View Synthesis of Dynamic Scenes

• 论文/paper：https://arxiv.org/abs/2108.13408
• 代码/code：https://cseweb.ucsd.edu//~viscomp/projects/ICCV21Deep/

Cross-category Video Highlight Detection via Set-based Learning

• 论文/paper：https://arxiv.org/abs/2108.11770
• 代码/code： https://github.com/ChrisAllenMing/Cross_Category_Video_Highlight

Overfitting the Data: Compact Neural Video Delivery via Content-aware Feature Modulation

• 论文/paper：https://arxiv.org/abs/2108.08202
• 代码/code： https://github.com/Anonymous-iccv2021-paper3163/CaFM-Pytorch

Sparse to Dense Motion Transfer for Face Image Animation

• 论文/paper：https://arxiv.org/abs/2109.00471
• 代码/code：None

SlowFast Rolling-Unrolling LSTMs for Action Anticipation in Egocentric Videos

• 论文/paper：https://arxiv.org/abs/2109.00829
• 代码/code：None

4D-Net for Learned Multi-Modal Alignment

• 论文/paper：https://arxiv.org/abs/2109.01066
• 代码/code：None

The Power of Points for Modeling Humans in Clothing

• 论文/paper：https://arxiv.org/abs/2109.01137
• 代码/code：None

The Functional Correspondence Problem

• 论文/paper：https://arxiv.org/abs/2109.01097
• 代码/code：None

On the Limits of Pseudo Ground Truth in Visual Camera Re-localisation

• 论文/paper：https://arxiv.org/abs/2109.00524
• 代码/code：None

Towards Learning Spatially Discriminative Feature Representations

• 论文/paper：https://arxiv.org/abs/2109.01359
• 代码/code：None

Learning Fast Sample Re-weighting Without Reward Data

• 论文/paper：https://arxiv.org/abs/2109.03216
• 代码/code：https://github.com/google-research/google-research/tree/master/ieg

CTRL-C: Camera calibration TRansformer with Line-Classification

• 论文/paper：https://arxiv.org/abs/2109.02259
• 代码/code：None

PR-Net: Preference Reasoning for Personalized Video Highlight Detection

• 论文/paper：https://arxiv.org/abs/2109.01799
• 代码/code：None

Dual Transfer Learning for Event-based End-task Prediction via Pluggable Event to Image Translation

• 论文/paper：https://arxiv.org/abs/2109.01801
• 代码/code：None

Learning to Generate Scene Graph from Natural Language Supervision

• 论文/paper：https://arxiv.org/abs/2109.02227
• 代码/code：https://github.com/YiwuZhong/SGG_from_NLS

Parsing Table Structures in the Wild

• 论文/paper：https://arxiv.org/abs/2109.02199
• 代码/code：None

Hierarchical Object-to-Zone Graph for Object Navigation

• 论文/paper：https://arxiv.org/abs/2109.02066
• 代码/code：None

Square Root Marginalization for Sliding-Window Bundle Adjustment

• 论文/paper：https://arxiv.org/abs/2109.02182
• 代码/code：None

YouRefIt: Embodied Reference Understanding with Language and Gesture

• 论文/paper：https://arxiv.org/abs/2109.03413
• 代码/code：None

Deep Hough Voting for Robust Global Registration

• 论文/paper：https://arxiv.org/abs/2109.04310
• 代码/code：None

IICNet: A Generic Framework for Reversible Image Conversion

• 论文/paper：https://arxiv.org/abs/2109.04242
• 代码/code：https://github.com/felixcheng97/IICNet

Estimating Leaf Water Content using Remotely Sensed Hyperspectral Data

• 论文/paper：https://arxiv.org/abs/2109.02250
• 代码/code：None

What Matters for Ad-hoc Video Search? A Large-scale Evaluation on TRECVID

• 论文/paper：https://arxiv.org/abs/2109.01774
• 代码/code：None

Shape-Biased Domain Generalization via Shock Graph Embeddings

• 论文/paper：https://arxiv.org/abs/2109.05671
• 代码/code：None

Explain Me the Painting: Multi-Topic Knowledgeable Art Description Generation

• 论文/paper：https://arxiv.org/abs/2109.05743
• 代码/code：None

Learning Indoor Inverse Rendering with 3D Spatially-Varying Lighting（Oral）

• 论文/paper：https://arxiv.org/abs/2109.06061
• 代码/code：None

Multiresolution Deep Implicit Functions for 3D Shape Representation

• 论文/paper：https://arxiv.org/abs/2109.05591
• 代码/code：None

Image Shape Manipulation from a Single Augmented Training Sample （Oral）

• 论文/paper：https://arxiv.org/abs/2109.06151
• 代码/code：None

ZFlow: Gated Appearance Flow-based Virtual Try-on with 3D Priors

• 论文/paper：https://arxiv.org/abs/2109.07001
• 代码/code：None

Contact-Aware Retargeting of Skinned Motion

• 论文/paper：https://arxiv.org/abs/2109.07431
• 代码/code：None

DisUnknown: Distilling Unknown Factors for Disentanglement Learning

• 论文/paper：https://arxiv.org/abs/2109.08090
• 代码/code：https://github.com/stormraiser/disunknown

FSER: Deep Convolutional Neural Networks for Speech Emotion Recognition

• 论文/paper：https://arxiv.org/abs/2109.07916
• 代码/code：None

A Pathology Deep Learning System Capable of Triage of Melanoma Specimens Utilizing Dermatopathologist Consensus as Ground Truth

• 论文/paper：https://arxiv.org/abs/2109.07554
• 代码/code：None

PIRenderer: Controllable Portrait Image Generation via Semantic Neural Rendering

• 论文/paper：https://arxiv.org/abs/2109.08379
• 代码/code：https://github.com/RenYurui/PIRender

The First Vision For Vitals (V4V) Challenge for Non-Contact Video-Based Physiological Estimation

• 论文/paper：https://arxiv.org/abs/2109.10471 | Dataset and Challenge
• 代码/code：None

FaceEraser: Removing Facial Parts for Augmented Reality

• 论文/paper：https://arxiv.org/abs/2109.10760
• 代码/code：None

S3VAADA: Submodular Subset Selection for Virtual Adversarial Active Domain Adaptation

• 论文/paper：https://arxiv.org/abs/2109.08901
• 代码/code：None

JEM++: Improved Techniques for Training JEM

• 论文/paper：https://arxiv.org/abs/2109.09032
• 代码/code：https://github.com/sndnyang/JEMPP

Rational Polynomial Camera Model Warping for Deep Learning Based Satellite Multi-View Stereo Matching

• 论文/paper：https://arxiv.org/abs/2109.11121
• 代码/code：https://github.com/WHU-GPCV/SatMVS

Long Short View Feature Decomposition via Contrastive Video Representation Learning

• 论文/paper：https://arxiv.org/abs/2109.11593
• 代码/code：None

Visual Scene Graphs for Audio Source Separation

• 论文/paper：https://arxiv.org/abs/2109.11955
• 代码/code：None

Meta-Aggregator: Learning to Aggregate for 1-bit Graph Neural Networks

• 论文/paper：https://arxiv.org/abs/2109.12872
• 代码/code：None

Modelling Neighbor Relation in Joint Space-Time Graph for Video Correspondence Learning

• 论文/paper：https://arxiv.org/abs/2109.13499
• 代码/code：None

Meta Learning on a Sequence of Imbalanced Domains with Difficulty Awareness

• 论文/paper：https://arxiv.org/abs/2109.14120
• 代码/code：None

Sensor-Guided Optical Flow

• 论文/paper：https://arxiv.org/abs/2109.15321
• 代码/code：None

CrossCLR: Cross-modal Contrastive Learning For Multi-modal Video Representations

• 论文/paper：https://arxiv.org/abs/2109.14910
• 代码/code：None

Video Autoencoder: self-supervised disentanglement of static 3D structure and motion

• 论文/paper：https://arxiv.org/abs/2110.02951
• 代码/code：None

Topologically Consistent Multi-View Face Inference Using Volumetric Sampling

• 论文/paper：https://arxiv.org/abs/2110.02948
• 代码/code：https://tianyeli.github.io/tofu

Extensions of Karger's Algorithm: Why They Fail in Theory and How They Are Useful in Practice （Oral）

• 论文/paper：https://arxiv.org/abs/2110.02750
• 代码/code：None

HighlightMe: Detecting Highlights from Human-Centric Videos

• 论文/paper：https://arxiv.org/abs/2110.01774
• 代码/code：None

How You Move Your Head Tells What You Do: Self-supervised Video Representation Learning with Egocentric Cameras and IMU Sensors

• 论文/paper：https://arxiv.org/abs/2110.01680
• 代码/code：None

Structured Bird's-Eye-View Traffic Scene Understanding from Onboard Images

• 论文/paper：https://arxiv.org/abs/2110.01997
• 代码/code：https://github.com/ybarancan/STSU

Waypoint Models for Instruction-guided Navigation in Continuous Environments

• 论文/paper：https://arxiv.org/abs/2110.02207
• 代码/code：None

Procedure Planning in Instructional Videosvia Contextual Modeling and Model-based Policy Learning （Oral）

• 论文/paper：https://arxiv.org/abs/2110.01770
• 代码/code：None

De-rendering Stylized Texts

• 论文/paper：https://arxiv.org/abs/2110.01890
• 代码/code：https://github.com/CyberAgentAILab/derendering-text

Spatio-Temporal Video Representation Learning for AI Based Video Playback Style Prediction

• 论文/paper：https://arxiv.org/abs/2110.01015
• 代码/code：None

Keypoint Communities

• 论文/paper：https://arxiv.org/abs/2110.00988
• 代码/code：None

Calibrating Concepts and Operations: Towards Symbolic Reasoning on Real Images

• 论文/paper：https://arxiv.org/abs/2110.00519
• 代码/code：https://github.com/Lizw14/CaliCO

A Hierarchical Variational Neural Uncertainty Model for Stochastic Video Prediction (Oral)

• 论文/paper：https://arxiv.org/abs/2110.03446
• 代码/code：None

2nd Place Solution to Google Landmark Retrieval 2021

• 论文/paper：https://arxiv.org/abs/2110.04294
• 代码/code：https://github.com/WesleyZhang1991/Google_Landmark_Retrieval_2021_2nd_Place_Solution

Neural Strokes: Stylized Line Drawing of 3D Shapes

• 论文/paper：https://arxiv.org/abs/2110.03900
• 代码/code：https://github.com/DifanLiu/NeuralStrokes

Learning Realistic Human Reposing using Cyclic Self-Supervision with 3D Shape, Pose, and Appearance Consistency

• 论文/paper：https://arxiv.org/abs/2110.05458
• 代码/code：None

Pano-AVQA: Grounded Audio-Visual Question Answering on 360∘ Videos

• 论文/paper：https://arxiv.org/abs/2110.05122
• 代码/code：None

Omnidata: A Scalable Pipeline for Making Multi-Task Mid-Level Vision Datasets from 3D Scans

• 论文/paper：https://arxiv.org/abs/2110.04994
• 代码/code：None

BuildingNet: Learning to Label 3D Buildings (oral)

• 论文/paper：https://arxiv.org/abs/2110.04955
• 代码/code：None

SOMA: Solving Optical Marker-Based MoCap Automatically

• 论文/paper：https://arxiv.org/abs/2110.04431
• 代码/code：None

Topic Scene Graph Generation by Attention Distillation from Caption

• 论文/paper：https://arxiv.org/abs/2110.05731
• 代码/code：None

Winning the ICCV'2021 VALUE Challenge: Task-aware Ensemble and Transfer Learning with Visual Concepts

• 论文/paper：https://arxiv.org/abs/2110.06476
• 代码/code：None

Understanding of Emotion Perception from Art

• 论文/paper：https://arxiv.org/abs/2110.06486
• 代码/code：None

Nuisance-Label Supervision: Robustness Improvement by Free Labels

• 论文/paper：https://arxiv.org/abs/2110.07118
• 代码/code：None

Simple Baseline for Single Human Motion Forecasting

• 论文/paper：https://arxiv.org/abs/2110.07495
• 代码/code：None

PixelPyramids: Exact Inference Models from Lossless Image Pyramids

• 论文/paper：https://arxiv.org/abs/2110.08787
• 代码/code：None

返回目录/back