Computer vision - ECCV 2022 . Part XXXI : 17th European Conference, Tel Aviv, Israel, October 23-27, 2022 : proceedings / / Shai Avidan [and four others]

• Pubblicazione: Cham, Switzerland : , : Springer, , [2022]; ©2022
• Descrizione fisica: 1 online resource (810 pages)
• Disciplina: 006.4
• Soggetto topico: Pattern recognition systems; Computer vision
• Persona (resp. second.): AvidanShai
• Nota di contenuto: Intro -- Foreword -- Preface -- Organization -- Contents - Part XXXI -- GOCA: Guided Online Cluster Assignment for Self-supervised Video Representation Learning -- 1 Introduction -- 2 Related Works -- 3 Method -- 3.1 Preliminaries -- 3.2 Guided Online Cluster Assignment (GOCA) -- 3.3 Prototype Regularization -- 3.4 Training Procedure -- 4 Experimental Results -- 4.1 Ablation Study -- 4.2 Retrieval Results -- 4.3 Cluster Analysis -- 4.4 Classification Tasks -- 5 Conclusion -- References -- Constrained Mean Shift Using Distant yet Related Neighbors for Representation Learning -- 1 Introduction -- 2 Method -- 2.1 Self-supervised Setting -- 2.2 Supervised Setting -- 2.3 Semi-supervised Setting -- 3 Experiments -- 3.1 Self-supervised Learning (CMSFself) -- 3.2 Supervised Learning -- 3.3 Semi-supervised Learning -- 4 Related Work -- 5 Conclusion -- References -- Revisiting the Critical Factors of Augmentation-Invariant Representation Learning -- 1 Introduction -- 2 Related Work -- 3 Experimental Setup -- 3.1 Framework -- 3.2 Pre-training and Evaluation -- 4 Experiments and Analyses -- 4.1 What Matters in Linear Evaluations? -- 4.2 How to Improve Transfer Performances? -- 4.3 What Is the Impact of Asymmetric Network Structure? -- 5 Conclusion -- References -- CA-SSL: Class-Agnostic Semi-Supervised Learning for Detection and Segmentation -- 1 Introduction -- 2 Related Work -- 3 Methodology -- 3.1 Training Stages -- 3.2 Switching from Objects to Entities -- 3.3 Base Detector -- 4 Experiments -- 4.1 Experimental Results -- 5 Conclusion -- References -- Dual Adaptive Transformations for Weakly Supervised Point Cloud Segmentation -- 1 Introduction -- 2 Related Work -- 2.1 Deep Learning on Point Clouds -- 2.2 Weakly Supervised Point Cloud Segmentation -- 2.3 Consistency-Based Semi-supervised Learning -- 3 Methods -- 3.1 Segmentation Module. ; 3.2 Local Adaptive Perturbation Module -- 3.3 Regional Adaptive Deformation Module -- 3.4 Training Losses -- 4 Experiments and Results -- 4.1 Implementation Details -- 4.2 Evaluations on S3DIS Dataset -- 4.3 Evaluations on ScanNet-v2 Dataset -- 4.4 Qualitative Results -- 5 Conclusion -- References -- Semantic-Aware Fine-Grained Correspondence -- 1 Introduction -- 2 Related Work -- 3 Method -- 3.1 Background -- 3.2 Semantic Correspondence Learning -- 3.3 Fine-Grained Correspondence Learning -- 3.4 Fusion of Correspondence Signals -- 3.5 Implementation Details -- 4 Experiments -- 4.1 Experimental Settings -- 4.2 Comparison with State-of-the-Art -- 4.3 Visualization -- 4.4 Ablative Analysis -- 5 Conclusion and Discussions -- References -- Self-Supervised Classification Network -- 1 Introduction -- 2 Related Work -- 2.1 Self-Supervised Learning -- 2.2 Deep Unsupervised Clustering -- 3 Self-Classifier -- 4 Theoretical Analysis -- 5 Implementation Details -- 5.1 Architecture -- 5.2 Image Augmentations -- 5.3 Optimization -- 6 Results -- 6.1 Unsupervised Image Classification -- 6.2 Image Classification with Linear Models -- 6.3 Transfer Learning -- 6.4 Qualitative Results -- 7 Ablation Study -- 8 Comparative Analysis -- 9 Conclusions and Limitations -- References -- Data Invariants to Understand Unsupervised Out-of-Distribution Detection -- 1 Introduction -- 2 Related Works -- 3 Invariants for Unsupervised OOD -- 3.1 Formalization -- 3.2 The Mahalanobis Anomaly Detector -- 4 Experiments -- 4.1 Results -- 4.2 Importance of Data Invariants -- 5 Conclusion -- References -- Domain Invariant Masked Autoencoders for Self-supervised Learning from Multi-domains -- 1 Introduction -- 2 Related Work -- 2.1 Self-supervised Learning -- 2.2 Domain Generalization -- 3 Domain-Invariant Masked AutoEncoder -- 3.1 Cross-domain Reconstruction Framework. ; 3.2 Content Preserved Style-Mix -- 3.3 Content Encoder -- 3.4 Domain Specific Decoders -- 3.5 Objective Function -- 4 Experiment -- 4.1 Experimental Setup -- 4.2 Experimental Results -- 4.3 Ablation Study -- 4.4 Visualization -- 5 Conclusions -- References -- Semi-supervised Object Detection via VC Learning -- 1 Introduction -- 2 Related Works -- 3 Methodology -- 3.1 Problem Definition -- 3.2 Virtual Category Learning -- 3.3 Potential Category Set -- 3.4 Localisation Loss -- 4 Experiments -- 4.1 Datasets and Evaluation Protocol -- 4.2 Implementation -- 4.3 Performance -- 4.4 Analysis and Ablation Study -- 5 Discussion -- 6 Conclusion -- References -- Completely Self-supervised Crowd Counting via Distribution Matching -- 1 Introduction -- 2 Related Work -- 3 Our Approach -- 3.1 Natural Crowds and Density Distribution -- 3.2 Stage 1: Learning Features with Self-supervision -- 3.3 Stage 2: Sinkhorn Training -- 4 Experiments and Analysis -- 4.1 Crowd Datasets -- 4.2 Performance with Limited Data -- 4.3 CSS-CCNN in True Practical Setting -- 4.4 Analysis of the Prior Distribution -- 4.5 Sensitivity Analysis for the Crowd Parameter -- 5 Conclusions -- References -- Coarse-To-Fine Incremental Few-Shot Learning -- 1 Introduction -- 2 Related Work -- 3 A New Problem C2FSCIL -- 4 A Simple Approach Knowe -- 4.1 Learning Embedding-Weights Contrastively -- 4.2 Freezing Memorized Classifier-Weights -- 4.3 Normalizing Classifier-Weights -- 5 Theoretical Analysis of Knowe for Stability-Plasticity -- 6 Experiments -- 6.1 New Overall Performance Measures -- 6.2 Datasets and Results -- 6.3 Implementation Details -- 6.4 Ablation Study -- 6.5 Performance Comparison and Analysis -- 7 Conclusion -- References -- Learning Unbiased Transferability for Domain Adaptation by Uncertainty Modeling -- 1 Introduction -- 2 Related Works -- 3 Methodology -- 3.1 Preliminaries. ; 3.2 Unbiased Transferability Estimation -- 3.3 Unbiased Domain Adaptation -- 4 Experiments -- 4.1 Datasets and Protocols -- 4.2 Experimental Results -- 4.3 Further Analysis and Discussion -- 5 Conclusion -- References -- Learn2Augment: Learning to Composite Videos for Data Augmentation in Action Recognition -- 1 Introduction -- 2 Related Work -- 3 Learn2Augment -- 3.1 Selector -- 3.2 Semantic Matching (SM) -- 3.3 Video Compositing (VC) -- 4 Optimization of Learn2Augment -- 4.1 Training the Selector -- 4.2 Training the Classifier -- 5 Experiments -- 5.1 Experimental Details -- 5.2 Architectural Changes for Different Settings -- 5.3 Ablation Study -- 5.4 Augmenting in the Semi-supervised Setting -- 5.5 Augmenting in the Few-shot Setting -- 5.6 Augmenting the Full Training Set -- 6 Limitations and Future Work -- 7 Conclusion -- References -- CYBORGS: Contrastively Bootstrapping Object Representations by Grounding in Segmentation -- 1 Introduction -- 2 Related Work -- 3 CYBORGS -- 3.1 CYBORGS Framework Abstraction -- 3.2 Mask-Dependent Contrastive Learning -- 3.3 Bootstrapping Segmentation Masks -- 3.4 Consistency as a Curriculum for Segmentation -- 4 Experiments -- 4.1 Experimental Settings -- 4.2 Main Results: Representation Learning -- 4.3 Segmentation Quality -- 4.4 Ablations and Discussion -- 5 Conclusion -- References -- PSS: Progressive Sample Selection for Open-World Visual Representation Learning -- 1 Introduction -- 2 Related Work -- 3 Methods -- 3.1 Problem Formalization -- 3.2 Progressive Sample Selection (PSS) Pipeline -- 3.3 Sample Selection -- 3.4 Clustering and Pseudo Labeling -- 4 Experiments -- 4.1 Evaluation Protocols -- 4.2 Implementation Details -- 4.3 Ablation Experiments -- 4.4 Image Retrieval Results -- 4.5 Face Verification Results -- 5 Conclusion -- References. ; Improving Self-supervised Lightweight Model Learning via Hard-Aware Metric Distillation -- 1 Introduction -- 2 Related Work -- 3 Proposed Method -- 3.1 Difference-Guided Positive and Negative Mining -- 3.2 Hard-Aware Metric Distillation -- 3.3 Learning Constraint -- 4 Experiments -- 4.1 Self-supervised Distillation on CIFAR100 -- 4.2 Self-supervised Distillation on ImageNet -- 4.3 Applicability on Supervised Distillation Framework -- 4.4 Analysis -- 5 Conclusion -- References -- Object Discovery via Contrastive Learning for Weakly Supervised Object Detection -- 1 Introduction -- 2 Related Work -- 2.1 Weakly Supervised Object Detection -- 2.2 Contrastive Learning -- 3 Background -- 3.1 Feature Extractor -- 3.2 Multiple Instance Learning Head -- 3.3 Refinement Head -- 4 Our Approach -- 4.1 Similarity Head -- 4.2 Sampling Strategy for Object Discovery -- 4.3 Object Discovery -- 4.4 Weakly Supervised Contrastive Loss -- 5 Experiments -- 5.1 Experiment Setting -- 5.2 Quantitative Results -- 5.3 Ablation Studies -- 5.4 Qualitative Results -- 6 Conclusion -- References -- Stochastic Consensus: Enhancing Semi-Supervised Learning with Consistency of Stochastic Classifiers -- 1 Introduction -- 2 Related Works -- 3 The Proposed Method -- 3.1 FixMatch -- 3.2 Our Method: Stochastic Consensus -- 3.3 Theoretical Analysis -- 4 Experiments -- 4.1 Setup -- 4.2 Ablation Studies and Learning Analyses -- 4.3 Comparison with the State-of-the-Art -- 5 Conclusion -- References -- DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model -- 1 Introduction -- 2 Backgrounds and Related Works -- 2.1 Deformable Image Registration -- 2.2 Denoising Diffusion Probabilistic Model -- 3 Proposed Method -- 3.1 Framework of DiffuseMorph -- 3.2 Loss Function -- 3.3 Image Registration Using DiffuseMorph -- 4 Experimental Results. ; 4.1 Results of Intra-subject 2D Face Image Registration.
• Titolo autorizzato: Computer Vision – ECCV 2022
• ISBN: 3-031-19821-2
• Formato: Materiale a stampa
• Livello bibliografico: Monografia
• Lingua di pubblicazione: Inglese
• Record Nr.: 996495568503316
• Lo trovi qui: Univ. di Salerno
• Serie: Lecture notes in computer science.