Awesome
ECCV2022-Paper-Code-Interpretation
ECCV2022 论文/代码/解读合集,极市团队整理
ECCV2022 最新论文分类
检索链接:https://arxiv.org/search/?query=ECCV2022&searchtype=all&source=header<br> 更新时间:2022年7月22日<br>
相关报道:ECCV 2022放榜了!1629篇论文中选,录用率不到20%
1.ECCV2022 接受论文/代码分方向整理(持续更新)
2.ECCV2022 oral
3.ECCV2022 论文解读汇总
update: <br>
2022/8/4 更新11篇<br>
<br><br>
<a name="eccv"/>ECCV2022 接受论文/代码分方向整理(持续更新)
目录
1. 检测
- 2D目标检测(2D Object Detection)
- 视频目标检测(Video Object Detection)
- 3D目标检测(3D Object Detection)
- 人物交互检测(HOI Detection)
- 伪装目标检测(Camouflaged Object Detection)
- 旋转目标检测(Rotation Object Detection)
- 显著性目标检测(Saliency Object Detection)
- 关键点检测(Keypoint Detection)
- 车道线检测(Lane Detection)
- 边缘检测(Edge Detection)
- 消失点检测(Vanishing Point Detection)
- 异常检测(Anomaly Detection)
2. 分割(Segmentation)
- 图像分割(Image Segmentation)
- 全景分割(Panoptic Segmentation)
- 语义分割(Semantic Segmentation)
- 实例分割(Instance Segmentation)
- 超像素(Superpixel)
- 视频目标分割(Video Object Segmentation)
- 抠图(Matting)
- 密集预测(Dense Prediction)
3. 图像处理(Image Processing)
- 超分辨率(Super Resolution)
- 图像复原/图像增强/图像重建(Image Restoration/Image Reconstruction)
- 图像去阴影/去反射(Image Shadow Removal/Image Reflection Removal)
- 图像去噪/去模糊/去雨去雾(Image Denoising/Deblurring/Dehazing)
- 图像编辑/图像修复(Image Edit/Image Inpainting)
- 图像翻译(Image Translation)
- 图像质量评估(Image Quality Assessment)
- 风格迁移(Style Transfer)
4. 视频处理(Video Processing)
- 视频编辑(Video Editing)
- 视频修复(Video Inpainting)
- 视频去模糊(Video Deblurring)
- 视频生成/视频合成(Video Generation/Video Synthesis)
- 视频超分(Video Super-Resolution)
5. 图像&视频检索/视频理解(Image&Video Retrieval/Video Understanding)
- 行为识别/行为识别/动作识别/检测/分割(Action/Activity Recognition)
- 行人重识别/检测(Re-Identification/Detection)
- 图像/视频字幕(Image/Video Caption)
- 视频理解(Video Understanding)
- 图像/视频检索(Image/Video Retrieval)
6. 估计(Estimation)
- 光流/运动估计(Flow/Motion Estimation)
- 深度估计(Depth Estimation)
- 人体解析/人体姿态估计(Human Parsing/Human Pose Estimation)
- 手势估计(Gesture Estimation)
7. 人脸(Face)
- 人脸识别/检测(Facial Recognition/Detection)
- 人脸生成/合成/重建/编辑(Face Generation/Face Synthesis/Face Reconstruction/Face Editing)
- 人脸伪造/反欺骗(Face Forgery/Face Anti-Spoofing)
8. 三维视觉(3D Vision)
9. 目标跟踪(Object Tracking)
10. 医学影像(Medical Imaging)
11. 文本检测/识别/理解(Text Detection/Recognition/Understanding)
12. 遥感图像(Remote Sensing Image)
13. GAN/生成式/对抗式(GAN/Generative/Adversarial)
14. 图像生成/图像合成(Image Generation/Image Synthesis)
15. 场景图(Scene Graph)
16. 视觉推理/视觉问答(Visual Reasoning/VQA)
17. 视觉预测(Vision-based Prediction)
18. 神经网络结构设计(Neural Network Structure Design)
19. 神经网络可解释性(Neural Network Interpretability)
20. 数据集(Dataset)
21. 数据处理(Data Processing)
22. 图像特征提取与匹配(Image feature extraction and matching)
23. 视觉表征学习(Visual Representation Learning)
24. 模型训练/泛化(Model Training/Generalization)
25. 模型压缩(Model Compression)
26. 模型评估(Model Evaluation)
27. 图像分类(Image Classification)
28. 图像计数(Image Counting)
29. 机器人(Robotic)
30. 半监督学习/弱监督学习/无监督学习/自监督学习(Self-supervised Learning/Semi-supervised Learning)
31. 多模态学习(Multi-Modal Learning)
32. 主动学习(Active Learning)
33. 小样本学习/零样本学习(Few-shot/Zero-shot Learning)
34. 持续学习(Continual Learning/Life-long Learning)
35. 迁移学习/domain/自适应(Transfer Learning/Domain Adaptation)
36. 度量学习(Metric Learning)
37. 对比学习(Contrastive Learning)
38. 增量学习(Incremental Learning)
39. 强化学习(Reinforcement Learning)
40. 元学习(Meta Learning)
41. 联邦学习(Federated Learning)
42. 模仿学习(Imitation Learning)
<br><br>
<br> <a name="detection"/>1. 检测
<br> <a name="IOD"/>2D目标检测(2D Object Detection)
[4] Multimodal Object Detection via Probabilistic Ensembling (基于概率集成的多模态目标检测) (Oral)<br>
[3] Point-to-Box Network for Accurate Object Detection via Single Point Supervision (通过单点监督实现精确目标检测的点对盒网络)<br> paper | code<br><br>
[2] You Should Look at All Objects (您应该查看所有物体)<br> paper | code<br><br>
[1] Adversarially-Aware Robust Object Detector (对抗性感知鲁棒目标检测器)(Oral))<br> paper | code<br><br>
<br> <a name="3DOD"/>3D目标检测(3D Object Detection)
[2] Densely Constrained Depth Estimator for Monocular 3D Object Detection (用于单目 3D 目标检测的密集约束深度估计器)<br> paper | code<br><br>
[1] Rethinking IoU-based Optimization for Single-stage 3D Object Detection (重新思考基于 IoU 的单阶段 3D 对象检测优化)<br> paper<br><br>
<br> <a name="VOD"/>视频目标检测(Video Object Detection)<br><br>
<br> <a name="HOI"/>人物交互检测(HOI Detection)
[2] Discovering Human-Object Interaction Concepts via Self-Compositional Learning (通过自组合学习发现人-物交互概念)<br>
paper | [code](https://github.com/zhihou7/scl; https://github.com/zhihou7/HOI-CL)<br><br>
[1] Towards Hard-Positive Query Mining for DETR-based Human-Object Interaction Detection (面向基于 DETR 的人机交互检测的硬性查询挖掘)<br> paper | code<br><br>
<br> <a name="SOD"/>显著性目标检测(Saliency Object Detection)<br><br>
[1] KD-SCFNet: Towards More Accurate and Efficient Salient Object Detection via Knowledge Distillation (KD-SCFNet:通过知识蒸馏实现更准确、更高效的显着目标检测)<br>
<br> <a name="COD"/>伪装目标检测(Camouflaged Object Detection)<br><br>
<br> <a name="ADI"/>图像异常检测/表面缺陷检测(Anomally Detection in Image)
[2] DSR -- A dual subspace re-projection network for surface anomaly detection (DSR——用于表面异常检测的双子空间重投影网络)<br>
[1] DICE: Leveraging Sparsification for Out-of-Distribution Detection (DICE:利用稀疏化进行分布外检测)<br> paper | code<br><br>
<br> <a name="EdgeDetection"/>边缘检测(Edge Detection)
<br><br>
<br> <a name="Segmentation"/>2. 分割(Segmentation)<br><br>
<br> <a name="ImageSegmentation"/>图像分割(Image Segmentation)
<br> <a name="InstanceSegmentation"/>实例分割(Instance Segmentation)
[3] In Defense of Online Models for Video Instance Segmentation (为视频实例分割的在线模型辩护) (Oral)<br> paper|code<br><br>
[2] Box-supervised Instance Segmentation with Level Set Evolution (具有水平集进化的框监督实例分割)<br> paper<br><br>
[1] OSFormer: One-Stage Camouflaged Instance Segmentation with Transformers (OSFormer:使用 Transformers 进行单阶段伪装实例分割)<br> paper | code<br><br>
<br> <a name="SemanticSegmentation"/>语义分割(Semantic Segmentation)
[1] 2DPASS: 2D Priors Assisted Semantic Segmentation on LiDAR Point Clouds (2DPASS:激光雷达点云上的二维先验辅助语义分割)<br> paper | code<br><br>
<br> <a name="VOS"/>视频目标分割(Video Object Segmentation)
[1] Learning Quality-aware Dynamic Memory for Video Object Segmentation (视频对象分割的学习质量感知动态内存)<br> paper | code<br><br>
<br> <a name="RIS"/>参考图像分割(Referring Image Segmentation)
<br> <a name="DensePrediction"/>密集预测(Dense Prediction)
<br><br>
<br> <a name="ImageProcessing"/>3. 图像处理(Image Processing)
<br> <a name="SuperResolution"/>超分辨率(Super Resolution)
[3] Learning Series-Parallel Lookup Tables for Efficient Image Super-Resolution (学习高效图像超分辨率的串并行查找表)<br>
[2] Efficient Meta-Tuning for Content-aware Neural Video Delivery (内容感知神经视频交付的高效元调整)<br> paper | code<br><br>
[1] Dynamic Dual Trainable Bounds for Ultra-low Precision Super-Resolution Networks (超低精度超分辨率网络的动态双可训练边界)<br> paper | code<br><br>
<br> <a name="ImageRestoration"/>图像复原/图像增强/图像重建(Image Restoration/Image Reconstruction)
[9] Unsupervised Night Image Enhancement: When Layer Decomposition Meets Light-Effects Suppression (无监督夜间图像增强:当层分解遇到光效抑制时)<br>
[8] Bringing Rolling Shutter Images Alive with Dual Reversed Distortion(通过双重反转失真使滚动快门图像重现) (Oral)<br> paper | code<br><br>
[7] Unsupervised Night Image Enhancement: When Layer Decomposition Meets Light-Effects Suppression (无监督夜间图像增强:当层分解遇到光效抑制时)<br> paper | code<br><br>
[6] Semantic-Sparse Colorization Network for Deep Exemplar-based Colorization (用于基于深度示例的着色的语义稀疏着色网络)<br> paper <br><br>
[5] Geometry-aware Single-image Full-body Human Relighting (几何感知单图像全身人体重新照明)<br> paper <br><br>
[4] Multi-Modal Masked Pre-Training for Monocular Panoramic Depth Completion (单目全景深度补全的多模态蒙面预训练)<br> paper <br><br>
[3] PanoFormer: Panorama Transformer for Indoor 360 Depth Estimation (PanoFormer:用于室内 360 深度估计的全景变压器)<br> paper <br><br>
[2] SESS: Saliency Enhancing with Scaling and Sliding (SESS:通过缩放和滑动增强显着性)<br> paper <br><br>
[1] RigNet: Repetitive Image Guided Network for Depth Completion (RigNet:用于深度补全的重复图像引导网络)<br> paper <br><br>
<br> <a name="ISR"/>图像去阴影/去反射(Image Shadow Removal/Image Reflection Removal)
[1] Deep Portrait Delighting (深度人像去光)<br>
paper<br><br>
<br> <a name="ImageDenoising"/>图像去噪(Image Denoising/Deblurring/Dehazing)
[3] Perceiving and Modeling Density is All You Need for Image Dehazing (感知和建模密度是图像去雾所需的全部) (Oral)<br> paper |code<br><br>
[2] Animation from Blur: Multi-modal Blur Decomposition with Motion Guidance (来自模糊的动画:具有运动引导的多模态模糊分解)<br> paper | code<br><br>
[1] Deep Semantic Statistics Matching (D2SM) Denoising Network (深度语义统计匹配(D2SM)去噪网络)<br> paper<br><br>
<br> <a name="ImageOutpainting"/>图像外推(Image Outpainting)
[1] Outpainting by Queries (通过查询进行外推)<br> paper | code<br><br>
<br> <a name="StyleTransfer"/>风格迁移(Style Transfer)
[1] CCPL: Contrastive Coherence Preserving Loss for Versatile Style Transfer (CCPL:通用风格迁移的对比相干性保留损失) (Oral)<br> paper | code<br><br>
<br><br>
<br> <a name="VideoProcessing"/>4. 视频处理(Video Processing)
<br> <a name="VideoEditing"/>视频编辑(Video Editing)
[3] AlphaVC: High-Performance and Efficient Learned Video Compression (AlphaVC:高性能和高效的学习视频压缩)<br>
paper<br>
<br>[2] Improving the Perceptual Quality of 2D Animation Interpolation (提高二维动画插值的感知质量)<br> paper | code<br><br>
[1] Real-Time Intermediate Flow Estimation for Video Frame Interpolation(视频帧插值的实时中间流估计)<br> paper | code<br><br>
<br> <a name="VideoInpainting"/>视频修复(Video Inpainting)
[1] Error Compensation Framework for Flow-Guided Video Inpainting (流引导视频修复的误差补偿框架)<br> paper<br><br>
<br> <a name="VideoDeblurring"/>视频去模糊(Video Deblurring)
[2] Event-guided Deblurring of Unknown Exposure Time Videos (未知曝光时间视频的事件引导去模糊) (Oral)<br>
paper<br><br>
[1] Efficient Video Deblurring Guided by Motion Magnitude (由运动幅度引导的高效视频去模糊)<br>
<br><br>
<br> <a name="ImageRetrieval"/>5. 图像&视频检索/视频理解(Image&Video Retrieval/Video Understanding)
<br> <a name="ActionRecognition"/>行为识别/行为识别/动作识别/检测/分割(Action/Activity Recognition)
[4] GaitEdge: Beyond Plain End-to-end Gait Recognition for Better Practicality (GaitEdge:超越普通的端到端步态识别,提高实用性)<br> paper | code<br><br>
[3] Collaborating Domain-shared and Target-specific Feature Clustering for Cross-domain 3D Action Recognition (用于跨域 3D 动作识别的协作域共享和特定于目标的特征聚类)<br> paper | code<br><br>
[2] ReAct: Temporal Action Detection with Relational Queries (ReAct:使用关系查询的时间动作检测)<br> paper | code<br><br>
[1] Hunting Group Clues with Transformers for Social Group Activity Recognition (用Transformers寻找群体线索用于社会群体活动识别)<br> paper <br><br>
<br> <a name="Re-Identification"/>行人重识别/检测(Re-Identification/Detection)
[1] PASS: Part-Aware Self-Supervised Pre-Training for Person Re-Identification(PASS:用于人员重新识别的部分感知自我监督预训练)<br> paper | code<br><br>
<br> <a name="VideoCaption"/>图像/视频字幕(Image/Video Caption)
<br> <a name="VideoUnderstanding"/>视频理解(Video Understanding)
[1] GraphVid: It Only Takes a Few Nodes to Understand a Video (GraphVid:只需几个节点即可理解视频) (Oral)<br> paper <br><br>
<br> <a name="VideoRetrieval"/>图像/视频检索(Image/Video Retrieval)
[6] Can Shuffling Video Benefit Temporal Bias Problem: A Novel Training Framework for Temporal Grounding (打乱的视频是否有益于时间偏差问题:一种新的时间接地训练框架)<br>
[5] Feature Representation Learning for Unsupervised Cross-domain Image Retrieval (无监督跨域图像检索的特征表示学习)<br> paper | code<br><br>
[4] LocVTP: Video-Text Pre-training for Temporal Localization (LocVTP:时间定位的视频文本预训练)<br> paper | code<br><br>
[3] Deep Hash Distillation for Image Retrieval (用于图像检索的深度哈希蒸馏)<br> paper | code<br><br>
[2] TS2-Net: Token Shift and Selection Transformer for Text-Video Retrieval (TS2-Net:用于文本视频检索的令牌移位和选择转换器)<br> paper | code<br><br>
[1] Lightweight Attentional Feature Fusion: A New Baseline for Text-to-Video Retrieval (轻量级注意力特征融合:文本到视频检索的新基线)<br> paper<br><br>
<br><br>
<br> <a name="Estimation"/>6. 估计(Estimation)
<br> <a name="Flow/Pose/MotionEstimation"/>光流/运动估计(Flow/Motion Estimation)
[1] Deep 360∘ Optical Flow Estimation Based on Multi-Projection Fusion (基于多投影融合的深度360∘光流估计)<br>
paper<br><br>
<br> <a name="VisualLocalization"/>视觉定位/位姿估计(Visual Localization/Pose Estimation)
[4] Overlooked Poses Actually Make Sense: Distilling Privileged Knowledge for Human Motion Prediction (被忽视的姿势实际上是有意义的:为人体运动预测提炼特权知识)<br>
paper<br><br>
[3] 3D Interacting Hand Pose Estimation by Hand De-occlusion and Removal (通过手部去遮挡和移除的 3D 交互手部姿势估计)<br>
[2] Weakly Supervised Object Localization via Transformer with Implicit Spatial Calibration (基于隐式空间校准的 Transformer 的弱监督目标定位)<br> [paper] (https://arxiv.org/abs/2207.10447) | code<br><br>
[1] Category-Level 6D Object Pose and Size Estimation using Self-Supervised Deep Prior Deformation Networks (使用自监督深度先验变形网络的类别级 6D 对象姿势和大小估计)<br> paper | code<br><br>
<br> <a name="DepthEstimation"/>深度估计(Depth Estimation)
[1] Physical Attack on Monocular Depth Estimation with Optimal Adversarial Patches ((使用最优对抗补丁对单目深度估计进行物理攻击))<br> paper <br><br>
<br><br>
<br> <a name="Face"/>7. 人脸(Face)
<br> <a name="FacialRecognition"/>人脸识别/检测(Facial Recognition/Detection)
[1] Towards Racially Unbiased Skin Tone Estimation via Scene Disambiguation (通过场景消歧实现种族无偏肤色估计)<br>
<br> <a name="FacialDetection"/>人脸识别/检测(Facial Recognition/Detection)
[1] MoFaNeRF: Morphable Facial Neural Radiance Field (MoFaNeRF:可变形面部神经辐射场)<br>
<br> <a name="FaceAnti-Spoofing"/>人脸伪造/反欺骗(Face Forgery/Face Anti-Spoofing)
<br><br>
<br> <a name="3DVision"/>8. 三维视觉(3D Vision)
<br> <a name="3DReconstruction"/>三维重建(3D Reconstruction)
[1] DiffuStereo: High Quality Human Reconstruction via Diffusion-based Stereo Using Sparse Cameras (DiffuStereo:使用稀疏相机通过基于扩散的立体进行高质量人体重建)<br> paper<br><br>
<br> <a name="NeRF"/>场景重建/视图合成/新视角合成(Novel View Synthesis)
[1] Sem2NeRF: Converting Single-View Semantic Masks to Neural Radiance Fields (Sem2NeRF:将单视图语义掩码转换为神经辐射场)<br> paper | code<br><br>
<br><br>
<br> <a name="ObjectTracking"/>9. 目标跟踪(Object Tracking)
[2] Tracking Every Thing in the Wild (追踪野外的每一件事)<br>
paper<br><br>
[1] Towards Grand Unification of Object Tracking (迈向目标跟踪的大统一) (Oral)<br> paper | code<br><br>
<br><br>
<br> <a name="MedicalImaging"/>10. 医学影像(Medical Imaging)
<br><br>
<br> <a name="TDR"/>11. 文本检测/识别/理解(Text Detection/Recognition/Understanding)
[5] Toward Understanding WordArt: Corner-Guided Transformer for Scene Text Recognition (了解艺术字:用于场景文本识别的角引导转换器) (Oral)<br>
[4] Contextual Text Block Detection towards Scene Text Understanding (面向场景文本理解的上下文文本块检测)<br>
paper<br><br>
[3] PromptDet: Towards Open-vocabulary Detection using Uncurated Images (PromptDet:使用未经处理的图像进行开放词汇检测)<br> paper |code<br><br>
[2] End-to-End Video Text Spotting with Transformer (使用 Transformer 的端到端视频文本定位) (Oral)<br> paper | code<br><br>
[1] Dynamic Low-Resolution Distillation for Cost-Efficient End-to-End Text Spotting (用于经济高效的端到端文本定位的动态低分辨率蒸馏)<br> paper | code<br><br>
<br><br>
<br> <a name="RSI"/>12. 遥感图像(Remote Sensing Image)
<br><br>
<br> <a name="GAN"/>13. GAN/生成式/对抗式(GAN/Generative/Adversarial)
[7] Learning Energy-Based Models With Adversarial Training (通过对抗训练学习基于能量的模型)<br>
[6] Adaptive Image Transformations for Transfer-based Adversarial Attack (基于传输的对抗性攻击的自适应图像转换)<br> paper<br><br>
[5] Generative Multiplane Images: Making a 2D GAN 3D-Aware (生成多平面图像:让一个2D GAN变得3D感知)<br> paper | code<br><br>
[4] Eliminating Gradient Conflict in Reference-based Line-Art Colorization (消除基于参考的艺术线条着色中的梯度冲突)<br> paper | code<br><br>
[3] WaveGAN: Frequency-aware GAN for High-Fidelity Few-shot Image Generation (WaveGAN:用于高保真少镜头图像生成的频率感知 GAN)<br> paper | code <br><br>
[2] FakeCLR: Exploring Contrastive Learning for Solving Latent Discontinuity in Data-Efficient GANs (FakeCLR:探索对比学习以解决数据高效 GAN 中的潜在不连续性)<br> paper | code <br><br>
[1] UniCR: Universally Approximated Certified Robustness via Randomized Smoothing (UniCR:通过随机平滑获得普遍近似的认证鲁棒性)<br> paper <br><br>
<br><br>
<br> <a name="IGIS"/>14. 图像生成/图像合成(Image Generation/Image Synthesis)
[1] PixelFolder: An Efficient Progressive Pixel Synthesis Network for Image Generation (PixelFolder:用于图像生成的高效渐进式像素合成网络)<br>
<br><br>
<br> <a name="SG"/>15. 场景图(Scene Graph)
<br><br>
<br> <a name="VisualReasoning"/>16. 视觉推理/视觉问答(Visual Reasoning/VQA)
<br><br>
<br> <a name="Vision-basedPrediction"/>17. 视觉预测(Vision-based Prediction)
[1] D2-TPred: Discontinuous Dependency for Trajectory Prediction under Traffic Lights (D2-TPred:交通灯下轨迹预测的不连续依赖)<br> paper | code<br><br>
<br><br>
<br> <a name="NNS"/>18. 神经网络结构设计(Neural Network Structure Design)
<br> <a name="DNN"/>DNN
[1] Hardly Perceptible Trojan Attack against Neural Networks with Bit Flips (使用 Bit Flips 对神经网络进行难以察觉的特洛伊木马攻击)<br>
<br> <a name="CNN"/>CNN
[1] PalQuant: Accelerating High-precision Networks on Low-precision Accelerators (PalQuant:在低精度加速器上加速高精度网络)<br>
<br> <br> <a name="Transformer"/>Transformer
[5] Point Primitive Transformer for Long-Term 4D Point Cloud Video Understanding (用于长期 4D 点云视频理解的 Point Primitive Transformer)<br>
paper<br><br>
[4] Improving Vision Transformers by Revisiting High-frequency Components (通过重新审视高频组件来改进视觉变压器)<br>
[3] Transformer with Implicit Edges for Particle-based Physics Simulation (用于基于粒子的物理模拟的隐式边缘变压器)<br>
[2] ScalableViT: Rethinking the Context-oriented Generalization of Vision Transformer (ScalableViT:重新思考 Vision Transformer 面向上下文的泛化)<br> paper | code<br><br>
[1] Visual Prompt Tuning (视觉提示调整)<br> paper | code<br><br>
<br> <a name="GNN"/>图神经网络(GNN)
<br> <a name="NAS"/>神经网络架构搜索(NAS)
[3] ScaleNet: Searching for the Model to Scale (ScaleNet:搜索要扩展的模型)<br> paper | code<br><br>
[2] Ensemble Knowledge Guided Sub-network Search and Fine-tuning for Filter Pruning (集成知识引导的子网络搜索和过滤器修剪微调)<br> paper | code<br><br>
[1] EAGAN: Efficient Two-stage Evolutionary Architecture Search for GANs (EAGAN:GAN 的高效两阶段进化架构搜索)<br> paper | code<br><br>
<br> <a name=" MLP"/>MLP
<br><br>
<br> <a name="interpretability"/>19. 神经网络可解释性(Neural Network Interpretability)
<br><br>
<br> <a name="Dataset"/>20. 数据集(Dataset)
<br> <a name="DataProcessing"/>21. 数据处理(Data Processing)
<br> <a name="DataAugmentation"/>数据增广(Data Augmentation)
<br> <a name="BatchNormalization"/>归一化/正则化(Batch Normalization)
[1] Fine-grained Data Distribution Alignment for Post-Training Quantization (训练后量化的细粒度数据分布对齐) (Oral)<br> paper | code<br><br>
<br> <a name="ImageClustering"/>图像聚类(Image Clustering)
<br> <a name="ImageCompression"/>图像压缩(Image Compression)
[1] Content-Oriented Learned Image Compression (面向内容的学习图像压缩)<br>
paper<br><br>
<br><br>
<br> <a name="matching"/>22. 图像特征提取与匹配(Image feature extraction and matching)
[1] Unsupervised Deep Multi-Shape Matching (无监督深度多形状匹配)<br> paper<br><br>
<br><br>
<br> <a name="VisualRL"/>23. 视觉表征学习(Visual Representation Learning)
[1] Object-Compositional Neural Implicit Surfaces (对象组合神经隐式曲面)<br> paper | code<br><br>
<br><br>
<br> <a name="ModelTraining"/>24. 模型训练/泛化(Model Training/Generalization)
<br><br>
<br> <a name="NoisyLabel"/>噪声标签(Noisy Label)
[1] Learning with Noisy Labels by Efficient Transition Matrix Estimation to Combat Label Miscorrection (通过有效的转移矩阵估计学习噪声标签以对抗标签错误校正)<br> paper <br><br>
<br> <a name="Long-Tailed"/>长尾分布(Long-Tailed Distribution)
[2] Long-tailed Instance Segmentation using Gumbel Optimized Loss (使用 Gumbel 优化损失的长尾实例分割)<br>
[1] Identifying Hard Noise in Long-Tailed Sample Distribution (识别长尾样本分布中的硬噪声) (Oral)<br>
<br><br>
<br> <a name="ModelCompression"/>25. 模型压缩(Model Compression)
<br> <a name="KnowledgeDistillation"/>知识蒸馏(Knowledge Distillation)
[3] Prune Your Model Before Distill It (在蒸馏之前修剪你的模型)<br>
[2] Efficient One Pass Self-distillation with Zipf's Label Smoothing (使用 Zipf 的标签平滑实现高效的单程自蒸馏)<br>
[1] Knowledge Condensation Distillation (知识浓缩蒸馏)<br> paper | code<br><br>
<br> <a name="Pruning"/>剪枝(Pruning)
<br> <a name="Quantization"/>量化(Quantization)
<br><br>
<br> <a name="ModelEvaluation"/>26. 模型评估(Model Evaluation)
[1] Hierarchical Latent Structure for Multi-Modal Vehicle Trajectory Forecasting (多模式车辆轨迹预测的分层潜在结构)<br> paper | code<br><br>
<br><br>
<br> <a name="ImageClassification"/>27. 图像分类(Image Classification)
[1] Centrality and Consistency: Two-Stage Clean Samples Identification for Learning with Instance-Dependent Noisy Labels (中心性和一致性:使用实例相关的噪声标签进行学习的两阶段清洁样本识别)<br>
<br><br>
<br> <a name="CrowdCounting"/>28. 图像计数(Image Counting)
<br><br>
<br> <a name="Robotic"/>29. 机器人(Robotic)
<br><br>
<br> <a name="self-supervisedlearning"/>30. 半监督学习/弱监督学习/无监督学习/自监督学习(Self-supervised Learning/Semi-supervised Learning)
[8] Acknowledging the Unknown for Multi-label Learning with Single Positive Labels (用单个正标签承认未知的多标签学习)<br>
[7] W2N:Switching From Weak Supervision to Noisy Supervision for Object Detection (W2N:目标检测从弱监督切换到嘈杂监督)<br>
[6] CA-SSL: Class-Agnostic Semi-Supervised Learning for Detection and Segmentation (CA-SSL:用于检测和分割的与类别无关的半监督学习)<br> paper | code<br><br>
[5] FedX: Unsupervised Federated Learning with Cross Knowledge Distillation (FedX:具有交叉知识蒸馏的无监督联合学习)<br> paper<br><br>
[4] Synergistic Self-supervised and Quantization Learning (协同自监督和量化学习)<br> paper | code<br><br>
[3] Contrastive Deep Supervision (对比深度监督)<br> paper | code<br><br>
[2] Dense Teacher: Dense Pseudo-Labels for Semi-supervised Object Detection (稠密教师:用于半监督目标检测的稠密伪标签)<br> paper<br><br>
[1] Image Coding for Machines with Omnipotent Feature Learning (具有全能特征学习的机器的图像编码)<br> paper <br><br>
<br><br>
<br> <a name="MMLearning"/>31. 多模态学习/跨模态(Multi-Modal Learning/Cross-Modal Learning)
<br><br>
<br> <a name="Audio-VisualLearning"/>视听学习(Audio-visual Learning)<br>
<br> <a name="VLRL"/>视觉-语言(Vision-language)
[2] Language Matters: A Weakly Supervised Vision-Language Pre-training Approach for Scene Text Detection and Spotting (语言问题:用于场景文本检测和识别的弱监督视觉语言预训练方法) (Oral)<br>
paper<br><br>
[1] Contrastive Vision-Language Pre-training with Limited Resources (资源有限的对比视觉语言预训练)<br> paper | code<br><br>
<br> <a name="CML"/>跨模态(cross-modal)
[1] Cross-modal Prototype Driven Network for Radiology Report Generation (用于放射学报告生成的跨模式原型驱动网络)<br> paper | code<br><br>
<br><br>
<br> <a name="ActiveLearning"/>32. 主动学习(Active Learning)
<br> <a name="Few-shotLearning"/>33. 小样本学习/零样本学习(Few-shot/Zero-shot Learning)
[2] Worst Case Matters for Few-Shot Recognition (最坏情况对少数镜头识别很重要)<br>
[1] Learning Instance and Task-Aware Dynamic Kernels for Few Shot Learning (用于少数镜头学习的学习实例和任务感知动态内核)<br> paper <br><br>
<br> <a name="ContinualLearning"/>34. 持续学习(Continual Learning/Life-long Learning)
[2] Balancing Stability and Plasticity through Advanced Null Space in Continual Learning (通过持续学习中的高级零空间平衡稳定性和可塑性) (Oral)<br>
paper<br><br>
[1] Online Continual Learning with Contrastive Vision Transformer (使用对比视觉转换器进行在线持续学习)<br>
paper<br><br>
<br><br>
<br> <a name="domain"/>35. 迁移学习/domain/自适应(Transfer Learning/Domain Adaptation)
[2] Factorizing Knowledge in Neural Networks (在神经网络中分解知识)<br> paper | code<br><br>
[1] CycDA: Unsupervised Cycle Domain Adaptation from Image to Video (CycDA:从图像到视频的无监督循环域自适应)<br> paper <br><br>
<br><br>
<br> <a name="MetricLearning"/>36. 度量学习(Metric Learning)
<br> <a name="ContrastiveLearning"/>37. 对比学习(Contrastive Learning)
<br> <a name="IncrementalLearning"/>38. 增量学习(Incremental Learning)
<br> <a name="RL"/>39. 强化学习(Reinforcement Learning)
[1] Target-absent Human Attention (目标缺失——人类注意力缺失)<br> paper | code<br><br>
<br> <a name="MetaLearning"/>40. 元学习(Meta Learning)
<br> <a name="FederatedLearning"/>41. 联邦学习(Federated Learning)
<br> <a name="ImitationLearning"/>42. 模仿学习(Imitation Learning)
[1] Resolving Copycat Problems in Visual Imitation Learning via Residual Action Prediction (通过残差动作预测解决视觉模仿学习中的模仿问题)<br> paper<br><br>
<br> <a name="oral"/>ECCV2022 Oral
[15] Toward Understanding WordArt: Corner-Guided Transformer for Scene Text Recognition (了解艺术字:用于场景文本识别的角引导转换器) (Oral)<br>
[14] Balancing Stability and Plasticity through Advanced Null Space in Continual Learning (通过持续学习中的高级零空间平衡稳定性和可塑性) (Oral)<br>
paper<br><br>
[13] Event-guided Deblurring of Unknown Exposure Time Videos (未知曝光时间视频的事件引导去模糊) (Oral)<br>
paper<br><br>
[12] Language Matters: A Weakly Supervised Vision-Language Pre-training Approach for Scene Text Detection and Spotting (语言问题:用于场景文本检测和识别的弱监督视觉语言预训练方法) (Oral)<br>
paper<br><br>
[11] Multimodal Object Detection via Probabilistic Ensembling (基于概率集成的多模态目标检测) (Oral)<br>
[10] Identifying Hard Noise in Long-Tailed Sample Distribution (识别长尾样本分布中的硬噪声) (Oral)<br>
[9] In Defense of Online Models for Video Instance Segmentation (为视频实例分割的在线模型辩护) (Oral)<br> paper|code<br><br>
[8] Perceiving and Modeling Density is All You Need for Image Dehazing (感知和建模密度是图像去雾所需的全部) (Oral)<br> paper |code<br><br>
[7] Bringing Rolling Shutter Images Alive with Dual Reversed Distortion(通过双重反转失真使滚动快门图像重现) (Oral)<br> paper | code<br><br>
[6] End-to-End Video Text Spotting with Transformer(使用 Transformer 的端到端视频文本定位) (Oral)<br> paper | code<br><br>
[5] GraphVid: It Only Takes a Few Nodes to Understand a Video(GraphVid:只需几个节点即可理解视频) (Oral)<br> paper <br><br>
[4] CCPL: Contrastive Coherence Preserving Loss for Versatile Style Transfer(CCPL:用于通用风格迁移的对比相干性保留损失) (Oral)<br> paper | code<br><br>
[3] Fine-grained Data Distribution Alignment for Post-Training Quantization(训练后量化的细粒度数据分布对齐) (Oral)<br> paper | code<br><br>
[2] Adversarially-Aware Robust Object Detector(对抗性感知鲁棒目标检测器) (Oral))<br> paper | code<br><br>
[1] Towards Grand Unification of Object Tracking(迈向目标跟踪的大统一) (Oral)<br> paper | code<br><br>
<br> <a name="3"/>ECCV2022 论文解读汇总
【1】文字解读:ECCV 2022 Oral | Unicorn:迈向目标跟踪的大统一<br> 直播解读:极市直播丨严彬-Unicorn:走向目标跟踪的大一统(ECCV2022 Oral)<br>
【2】ECCV 2022 Oral | 无需微调即可泛化!RegAD:少样本异常检测新框架<br>
【3】ECCV 2022 | Poseur:你以为我是姿态估计,其实是目标检测<br>
【4】ECCV 2022 | 清华&腾讯AI Lab提出REALY: 重新思考3D人脸重建的评估方法<br>
【5】ECCV 2022 | AirDet: 无需微调的小样本目标检测方法<br>
【6】ECCV2022 | 重新思考单阶段3D目标检测中的IoU优化<br>
【7】ECCV 2022 | 通往数据高效的Transformer目标检测器<br>
【8】ECCV2022 | FPN错位对齐,实现高效半监督目标检测 (PseCo)<br>
【9】ECCV 2022 | SmoothNet:用神经网络代替平滑滤波器,不用重新训练才配叫“即插即用”<br>
【10】ECCV2022 Oral | 无需前置条件的自动着色算法<br>
【11】ECCV2022|何恺明团队开源ViTDet:只用普通ViT,不做分层设计也能搞定目标检测
【12】ECCV 2022 | Masked Generative Distillation: 适用于分类,检测,分割的生成式知识蒸馏
【13】ECCV 2022 | 多域长尾分布学习,不平衡域泛化问题研究
【14】ECCV2022 | DisCo: 提升轻量化模型在自监督学习中的效果
【15】ECCV2022 最新综述 | 面向大规模场景的小目标检测:综述和 benchmark
【16】ECCV2022 | 京东&北航&美团提出时序动作检测新框架 性能SOTA!
【17】ECCV2022 | 你没见过的《老友记》镜头,AI给补出来了
【18】ECCV2022 | 重新思考单阶段3D目标检测中的IoU优化
【19】ECCV2022 | FPN错位对齐,实现高效半监督目标检测 (PseCo)
【20】ECCV2022|CV核心特征分解用于批处理矩阵 | 中小型矩阵的批量高效(batch-efficient)特征分解
【21】ECCV22 | CMU提出首个快速知识蒸馏的视觉框架:80.1%精度,训练加速30%
【22】ECCV 22|首个360°全景定制的单目深度估计Transformer-PanoFormer
【23】ECCV22|单点监督目标检测!国科大提出P2BNet:标一个点就能训练出强有力的目标检测器
【24】ECCV22|重新思考视觉Transformer面向上下文的泛化!清华&字节提出ScalableViT
【25】ECCV 2022 Oral|CCPL: 一种通用的关联性保留损失函数实现通用风格迁移
【26】ECCV 2022 | 仅用全连接层处理视频数据,美图&NUS实现高效视频时空建模
【27】ECCV 2022|计算机视觉中的长尾分布问题还值得做吗
【28】ECCV 22|大数据的红利我吃定了!微软开源TinyViT :搞定小模型的预训练能力