Computer vision - ECCV 2022 . Part XIV : 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 (807 pages)
• Disciplina: 006.37
• Soggetto topico: Computer vision; Pattern recognition systems
• Persona (resp. second.): AvidanShai
• Nota di contenuto: Intro -- Foreword -- Preface -- Organization -- Contents - Part XIV -- Watermark Vaccine: Adversarial Attacks to Prevent Watermark Removal -- 1 Introduction -- 2 Related Work -- 3 Methodology -- 3.1 Preliminary -- 3.2 Disrupting Watermark Vaccine (DWV) -- 3.3 Inerasable Watermark Vaccine (IWV) -- 4 Experiments -- 4.1 Experimental Setups -- 4.2 Effectiveness of Watermark Vaccine -- 4.3 Universality of Watermark Vaccine -- 4.4 Transferability of Watermark Vaccine -- 4.5 Resistance to Image Processing Operations -- 5 Conclusions -- References -- Explaining Deepfake Detection by Analysing Image Matching -- 1 Introduction -- 2 Related Work -- 2.1 Deepfake Detection -- 2.2 Interpretability of DNNs -- 3 Algorithms -- 3.1 Artifact Representations for Deepfake Detection Models -- 3.2 Learning the Artifact Representations -- 3.3 Vulnerability of Artifact Representations to Video Compression -- 3.4 FST-Matching Deepfake Detection Model -- 4 Experiment -- 4.1 Implementation Details -- 4.2 Fairness of the Shapley Value -- 4.3 Verification of Hypotheses -- 4.4 FST-Matching Deepfake Detection Model -- 5 Conclusions -- References -- FrequencyLowCut Pooling - Plug and Play Against Catastrophic Overfitting -- 1 Introduction -- 1.1 Related Work -- 2 Preliminaries -- 2.1 Adversarial Training -- 2.2 Down-Sampling in CNNs -- 3 FrequencyLowCut Pooling -- 4 Experiments -- 4.1 Native Robustness of FLC Pooling -- 4.2 FLC Pooling for FGSM Training -- 4.3 Training Efficiency -- 4.4 Black Box Attacks -- 4.5 Corruption Robustness -- 4.6 Shift-Invariance -- 5 Discussion and Conclusions -- A Appendix -- A.1 Training Schedules -- A.2 ImageNet Training Efficiency -- A.3 Aliasing Free Down-Sampling -- A.4 Model Confidences -- A.5 AutoAttack Attack Structure -- A.6 Ablation Study: Additional Frequency Components -- References. ; TAFIM: Targeted Adversarial Attacks Against Facial Image Manipulations -- 1 Introduction -- 2 Related Work -- 3 Proposed Method -- 3.1 Method Overview -- 3.2 Methodology -- 4 Results -- 5 Conclusion -- References -- FingerprintNet: Synthesized Fingerprints for Generated Image Detection -- 1 Introduction -- 2 Related Work -- 2.1 Generated Image Detection -- 2.2 Advancement in Generative Models -- 3 Fingerprint Generator -- 3.1 Overall Architecture -- 3.2 Training Loss -- 3.3 Random Layer Selection -- 3.4 Multi-kernel Deconvolution Layer -- 3.5 Feature Blender -- 3.6 Fingerprint Generation -- 4 Generated Image Detector -- 4.1 Effect of Frequency-Level Input -- 4.2 Architecture of Detector -- 4.3 Training Method with Mixed Batch -- 5 Experimental Results -- 5.1 Dataset -- 5.2 Evaluation Metrics -- 5.3 Generalization Performance of Our Detector -- 5.4 Generalization for Recent Generative Models -- 5.5 Color Manipulation Performance -- 5.6 Visualization -- 6 Conclusion -- References -- Detecting Generated Images by Real Images -- 1 Introduction -- 2 Related Work -- 3 Method -- 3.1 Learned Noise Patterns (LNP) -- 3.2 LNP Amplitude Spectrum -- 3.3 LNP Phase Spectrum -- 3.4 LNP Network -- 4 Experiments -- 4.1 Datasets -- 4.2 Setup -- 4.3 Comparisons -- 4.4 Ablation Study -- 4.5 Robustness -- 5 Conclusions -- References -- An Information Theoretic Approach for Attention-Driven Face Forgery Detection -- 1 Introduction -- 2 Related Work -- 2.1 Forgery Face Manipulation -- 2.2 Face Forgery Detection -- 3 Proposed Method -- 3.1 Preliminaries -- 3.2 Overall Framework -- 3.3 Self-information Computation -- 3.4 Self-information Based Dual Attention -- 3.5 Self-information Aggregation -- 3.6 Loss Function -- 4 Experiment -- 4.1 Experimental Setup -- 4.2 Experimental Results -- 4.3 Ablation Study -- 4.4 Visualization and Analysis -- 5 Conclusion -- References. ; Exploring Disentangled Content Information for Face Forgery Detection -- 1 Introduction -- 2 Related Works -- 2.1 Forgery Detection -- 2.2 Disentangled Representation -- 3 Methods -- 3.1 Motivation -- 3.2 Basic Disentanglement Framework -- 3.3 Enhanced Independence of Disentangled Features -- 3.4 Overall Loss -- 4 Experiments -- 4.1 Experimental Setting -- 4.2 Evaluations -- 4.3 Ablation Study -- 4.4 Augmentation Study of Disentangled Features -- 4.5 Investigation of Intrinsic Content Bias -- 4.6 Visualization -- 5 Conclusion -- References -- RepMix: Representation Mixing for Robust Attribution of Synthesized Images -- 1 Introduction -- 2 Related Work -- 3 The Attribution88 Benchmark -- 4 Methodology -- 4.1 Representation Mixing (RepMix) Layer -- 4.2 Compound Loss -- 5 Experiments -- 5.1 Training Details -- 5.2 Baseline Comparison -- 5.3 Robustness Against Individual Perturbation -- 5.4 Generalization on Semantic and Perturbation -- 5.5 Robustness Against Adversarial Attacks -- 5.6 Ablation Study -- 5.7 Further Analysis -- 6 Conclusion -- References -- Totems: Physical Objects for Verifying Visual Integrity -- 1 Introduction -- 2 Related Work -- 3 The Totem Verification Framework -- 3.1 Physical Totems for Image Verification -- 3.2 The General Totem Framework -- 4 Method -- 4.1 Scene Reconstruction from Totem Views -- 4.2 Manipulation Detection -- 5 Results -- 5.1 Data Collection -- 5.2 Decoding the Scene from Totem Views -- 5.3 Potential Avenues for a More General Method -- 6 Conclusion -- References -- Dual-Stream Knowledge-Preserving Hashing for Unsupervised Video Retrieval -- 1 Introduction -- 2 Related Work -- 3 Method -- 3.1 Problem Definition -- 3.2 Network Overview -- 3.3 Dual-Stream Reconstruction Learning -- 3.4 Semantic Knowledge Preservation -- 3.5 Overall Learning -- 4 Experimental Results. ; 4.1 Datasets, Metrics and Implementation Details -- 4.2 Comparisons with State-of-the-Art (SOTA) Methods -- 4.3 Ablation Study -- 4.4 Further Analysis -- 5 Conclusion -- References -- PASS: Part-Aware Self-Supervised Pre-Training for Person Re-Identification -- 1 Introduction -- 2 Related Work -- 2.1 Self-supervised Learning -- 2.2 Person Re-identification -- 3 Method -- 3.1 Preliminaries -- 3.2 Part-Aware Self-supervised Pre-training -- 3.3 Fine-Tuning -- 4 Experiments -- 4.1 Implementation Details -- 4.2 Comparison with State-of-the-Art Methods -- 4.3 Ablation Studies -- 5 Conclusion -- References -- Adaptive Cross-domain Learning for Generalizable Person Re-identification -- 1 Introduction -- 2 Related Work -- 2.1 Domain Generalization -- 2.2 Domain Generalizable Person Re-identification -- 2.3 Dynamic Neural Networks -- 3 Adaptive Cross-domain Learning Framework -- 3.1 Overview -- 3.2 Cross-domain Embedding Block -- 3.3 Objective Function -- 3.4 Optimization Process -- 4 Experiments -- 4.1 Implementation Details -- 4.2 Datasets and Evaluation Settings -- 4.3 Comparison with the State-of-the-Arts -- 4.4 Ablation Study -- 4.5 Visualization -- 5 Discussion -- 6 Conclusion -- References -- Multi-query Video Retrieval -- 1 Introduction -- 2 Related Work -- 3 Multi-query Video Retrieval -- 3.1 Setting -- 3.2 Methods -- 3.3 Evaluation -- 4 Experiments -- 4.1 Architecture Backbones -- 4.2 Experimental Setup -- 4.3 Results -- 5 Conclusion -- References -- Hierarchical Average Precision Training for Pertinent Image Retrieval -- 1 Introduction -- 2 Related Work -- 2.1 Image Retrieval and Ranking -- 2.2 Hierarchical Predictions and Metrics -- 3 HAPPIER Model -- 3.1 Hierarchical Average Precision -- 3.2 Direct Optimization of H-AP -- 4 Experiments -- 4.1 Experimental Setup -- 4.2 Main Results -- 4.3 HAPPIER Analysis -- 4.4 Qualitative Study -- 5 Conclusion. ; References -- Learning Semantic Correspondence with Sparse Annotations -- 1 Introduction -- 2 Related Work -- 2.1 Semantic Correspondence -- 2.2 Teacher-Student Learning -- 3 Model Architecture -- 3.1 Efficient Spatial Context Encoder -- 3.2 Correlation Map Computation -- 3.3 Flow Formation and High-resolution Loss -- 4 Learning with Sparse Annotations -- 4.1 Sparse Label Densification via Teacher-Student Learning -- 4.2 High Quality Pseudo-label Generation -- 4.3 Variants of Teacher-Student Learning -- 5 Experiments -- 5.1 Implementation Details -- 5.2 Comparison with State-of-the-Art Methods -- 5.3 Ablation Study -- 6 Conclusion -- References -- Dynamically Transformed Instance Normalization Network for Generalizable Person Re-Identification -- 1 Introduction -- 2 Related Works -- 3 Proposed Methods -- 3.1 Preliminaries -- 3.2 Dynamically Transformed Instance Normalization -- 3.3 Multi-task Training Strategy -- 4 Experiments -- 4.1 Implementation Details and Evaluation Setting -- 4.2 Comparison with State-of-the-Art Methods -- 4.3 Ablation Studies -- 5 Conclusion -- References -- Domain Adaptive Person Search -- 1 Introduction -- 2 Related Work -- 2.1 Person Search -- 2.2 Domain Adaptation for Person ReID -- 2.3 Domain Adaptive Object Detection -- 3 Methodology -- 3.1 Framework Overview -- 3.2 Domain Alignment Module -- 3.3 Training on Unlabeled Target Domain -- 4 Experiment -- 4.1 Datasets and Evaluation Protocols -- 4.2 Implementation Details -- 4.3 Ablation Study -- 4.4 Comparison with State-of-the-Art Methods -- 4.5 Qualitative Results -- 5 Conclusions -- References -- TS2-Net: Token Shift and Selection Transformer for Text-Video Retrieval -- 1 Introduction -- 2 Related Work -- 2.1 Video Retrieval -- 2.2 Visual-Language Pre-training -- 2.3 Video Representation Learning -- 3 Method -- 3.1 Token Shift Transformer. ; 3.2 Token Selection Transformer.
• Titolo autorizzato: Computer Vision – ECCV 2022
• ISBN: 3-031-19781-X
• Formato: Materiale a stampa
• Livello bibliografico: Monografia
• Lingua di pubblicazione: Inglese
• Record Nr.: 9910620200003321
• Lo trovi qui: Univ. Federico II
• Serie: Lecture notes in computer science.