Computer vision - ECCV 2022 . Part XIII : 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 (804 pages)
• Disciplina: 006.37
• Soggetto topico: Computer vision; Pattern recognition systems - Mathematical models
• Persona (resp. second.): AvidanShai
• Nota di contenuto: Intro -- Foreword -- Preface -- Organization -- Contents - Part XIII -- AU-Aware 3D Face Reconstruction through Personalized AU-Specific Blendshape Learning -- 1 Introduction -- 2 Related Works -- 3 Proposed Method -- 3.1 AU-specific 3D Model Learning -- 3.2 AU-aware 3D Face Reconstruction -- 3.3 Training Loss -- 4 Experiments -- 4.1 Evaluation of 3D Reconstruction Error -- 4.2 Evaluation of 3D AU Detection Performance -- 5 Conclusion -- References -- BézierPalm: A Free Lunch for Palmprint Recognition -- 1 Introduction -- 2 Related Work -- 2.1 Palmprint Recognition -- 2.2 Data Synthesis for Deep Models -- 3 Methodology -- 3.1 Palmar Creases with Bèzier Curves -- 3.2 Parameters of Bèziers -- 3.3 Within-Identity Diversity -- 4 Experimental Settings -- 4.1 Datasets and Data Preprocessing -- 4.2 Open-Set Protocol -- 4.3 Closed-Set Protocol -- 5 Experimental Results -- 5.1 Implementation Details -- 5.2 Open-Set Palmprint Recognition -- 5.3 Closed-Set Palmprint Recognition -- 5.4 Cross-Dataset Validation -- 5.5 Palmprint Recognition at Million Scale -- 5.6 Palmprint Recognition with Limited Identities -- 5.7 Ablation Study -- 6 Conclusion -- References -- Adaptive Transformers for Robust Few-shot Cross-domain Face Anti-spoofing -- 1 Introduction -- 2 Related Work -- 3 Method -- 3.1 Vision Transformer -- 3.2 Ensemble Adapters -- 3.3 Feature-wise Transformation -- 3.4 Adaptive Transformer -- 4 Experiments -- 4.1 Experimental Setups -- 4.2 Cross-domain Performance -- 4.3 Ablation Study -- 4.4 Visualization -- 5 Conclusion -- References -- Face2Facerho: Real-Time High-Resolution One-Shot Face Reenactment -- 1 Introduction -- 2 Related Work -- 3 Method -- 3.1 3DMM Fitting -- 3.2 Rendering Module -- 3.3 Motion Module -- 3.4 Training and Inference -- 4 Evaluation -- 4.1 Comparisons -- 4.2 Ablation Study -- 4.3 Limitation -- 5 Mobile Device Deployment. ; 6 Conclusion -- References -- Towards Racially Unbiased Skin Tone Estimation via Scene Disambiguation -- 1 Introduction -- 2 Related Work -- 3 Dataset and Metrics for Quantifying Skin Tone Bias -- 4 Reasons Behind Racially Biased Albedo Estimations -- 5 Unbiased Estimation via Scene Disambiguation Cues -- 5.1 Model -- 5.2 TRUST Network -- 5.3 Implementation Details -- 6 Evaluation -- 6.1 Qualitative Results -- 6.2 Quantitative Comparisons -- 6.3 Ablation Studies -- 7 Conclusions -- References -- BoundaryFace: A Mining Framework with Noise Label Self-correction for Face Recognition -- 1 Introduction -- 2 Related Work -- 2.1 Margin-Based Softmax -- 2.2 Mining-Based Softmax -- 3 The Proposed Approach -- 3.1 Preliminary Knowledge -- 3.2 Label Self-correction -- 3.3 BoundaryFace -- 3.4 Discussions with SOTA Loss Functions -- 4 Experiments -- 4.1 Implementation Details -- 4.2 Hyper-parameters -- 4.3 Comparisons with SOTA Methods -- 5 Conclusions -- References -- Pre-training Strategies and Datasets for Facial Representation Learning -- 1 Introduction -- 2 Related Work -- 3 Method -- 3.1 Network Pre-training -- 3.2 Pre-training Datasets -- 3.3 Facial Task Adaptation -- 3.4 Few-Shot Learning-Based Evaluation -- 3.5 Self-distillation for Semi-supervised Learning -- 4 Ablation Studies -- 5 Main Results -- 5.1 Face Recognition -- 5.2 Facial Landmark Localization -- 5.3 Action Unit (AU) Intensity Estimation -- 5.4 Emotion Recognition -- 5.5 3D Face Reconstruction -- 6 Discussion and Conclusions -- References -- Look Both Ways: Self-supervising Driver Gaze Estimation and Road Scene Saliency -- 1 Introduction -- 2 Related Work -- 3 Look Both Ways (LBW) Dataset -- 3.1 Calibration and Pre-processing -- 3.2 Final Collection -- 4 Self-supervised Gaze -- 4.1 Gaze-Driven Saliency -- 4.2 Losses and Network Design -- 4.3 Implementation Details -- 5 Experiments. ; 5.1 Evaluation Metrics -- 5.2 Baselines -- 5.3 Quantitative Evaluation -- 5.4 Qualitative Evaluation -- 6 Discussion -- References -- MFIM: Megapixel Facial Identity Manipulation -- 1 Introduction -- 2 Related Work -- 3 MFIM: Megapixel Facial Identity Manipulation -- 3.1 Facial Attribute Encoder -- 3.2 Training Objectives -- 3.3 ID Mixing -- 4 Experiments -- 4.1 Experimental Settings -- 4.2 Comparison with the Baselines -- 4.3 Ablation Study of MFIM -- 4.4 ID Mixing -- 5 Conclusion -- References -- 3D Face Reconstruction with Dense Landmarks -- 1 Introduction -- 2 Related Work -- 3 Method -- 3.1 Landmark Prediction -- 3.2 3D Model Fitting -- 3.3 Implementation -- 4 Evaluation -- 4.1 Landmark Accuracy -- 4.2 3D Face Reconstruction -- 4.3 Facial Performance Capture -- 5 Limitations and Future Work -- References -- Emotion-aware Multi-view Contrastive Learning for Facial Emotion Recognition -- 1 Introduction -- 2 Related Work -- 3 Method -- 3.1 Overview -- 3.2 Preliminaries of CRL -- 3.3 Proposed Method: AVCE -- 3.4 Similarity Function Design -- 3.5 Feature Transformations for Self-supervision -- 3.6 Discriminative Learning -- 4 Experiments -- 4.1 Datasets -- 4.2 Configurations -- 4.3 Quantitative Analysis -- 4.4 Qualitative Analysis -- 4.5 Ablation Study -- 5 Discussion of Limitations -- 6 Conclusion -- References -- Order Learning Using Partially Ordered Data via Chainization -- 1 Introduction -- 2 Related Work -- 2.1 Order Learning -- 2.2 Linear Extension of Partial Order -- 3 Proposed Algorithm -- 3.1 Preliminary -- 3.2 Problem Definition -- 3.3 Chainization - Basics -- 3.4 Chainization - Applications -- 3.5 Rank Estimation -- 4 Experiments -- 4.1 Datasets -- 4.2 Metrics -- 4.3 Random Edge Case -- 4.4 Clique-Edgeless Case -- 4.5 Bipartite Case -- 5 Conclusions -- References -- Unsupervised High-Fidelity Facial Texture Generation and Reconstruction. ; 1 Introduction -- 2 Background and Related Efforts -- 2.1 Our Contribution -- 3 Unsupervised Learning of Facial Textures and Geometries -- 3.1 Transfer Learning -- 3.2 Pose and Illumination Invariant Textures -- 3.3 Recovering Corresponding Geometry via Two-Step Fitting -- 3.4 Unsupervised Training -- 4 Experimental Results -- 4.1 Face Generation -- 4.2 Facial Texture Reconstruction -- 4.3 Ablation Study -- 4.4 Quantitative Results -- 5 Discussion, Limitations, and Future Work -- References -- Multi-domain Learning for Updating Face Anti-spoofing Models -- 1 Introduction -- 2 Related Works -- 3 Proposed Method -- 3.1 FAS Models Study -- 3.2 Spoof Region Estimator -- 3.3 FAS-Wrapper Architecture -- 3.4 Training and Inference -- 4 FASMD Dataset -- 5 Experimental Evaluations -- 5.1 Experiment Setup -- 5.2 Main Results -- 5.3 Adaptability Analysis -- 5.4 Algorithm Analysis -- 5.5 Cross-Dataset Study -- 6 Conclusion -- References -- Towards Metrical Reconstruction of Human Faces -- 1 Introduction -- 2 Related Work -- 3 Metrical Face Shape Prediction -- 4 Face Tracking -- 5 Dataset Unification -- 6 Results -- 6.1 Face Shape Estimation -- 6.2 Limitations -- 7 Discussion and Conclusion -- References -- Discover and Mitigate Unknown Biases with Debiasing Alternate Networks -- 1 Introduction -- 2 Related Work -- 3 Method -- 3.1 Unknown Bias Discovery -- 3.2 Unknown Bias Mitigation by Reweighing -- 3.3 Full Model -- 4 Experiment -- 4.1 Experiment on Multi-Color MNIST -- 4.2 Experiments on Face Image Dataset -- 4.3 Experiments on Other Image Domains -- 5 Conclusion -- References -- Unsupervised and Semi-supervised Bias Benchmarking in Face Recognition -- 1 Introduction -- 2 Related Work -- 3 Semi-supervised Bias Evaluation Methodology -- 3.1 Semi-supervised Performance Evaluation for Face Verification -- 3.2 Bias Evaluation in Face Recognition Systems. ; 3.3 Parametric Conditional Distributions and Priors -- 3.4 Bayesian Inference Strategy -- 4 Experiments -- 4.1 Face Embedding Model Training for Bias Analysis -- 4.2 Result and Analysis -- 5 Discussion and Conclusions -- References -- Towards Efficient Adversarial Training on Vision Transformers -- 1 Introduction -- 2 Relate Works -- 3 Fast Adversarial Training on Vision Transformers -- 4 Efficient Adversarial Training on Vision Transformers -- 4.1 Computation Intensity of ViTs -- 4.2 Dropping Patch: The Flexibility of Self-attention -- 4.3 Attention-Guided Adversarial Training -- 5 Experiments -- 5.1 Experimental Setup -- 5.2 Improved Efficiency of Adversarial Training on ImageNet -- 5.3 Ablation Study -- 6 Conclusions -- References -- MIME: Minority Inclusion for Majority Group Enhancement of AI Performance*-4pt -- 1 Introduction -- 1.1 Contributions -- 1.2 Outline of Theoretical Scope -- 2 Related Work -- 3 Statistical Origins of the MIME Effect -- 4 Verifying MIME Theory on Real Tasks -- 4.1 Verifying Assumptions -- 4.2 MIME Effect Across Six, Real Datasets -- 5 Discussion -- References -- Studying Bias in GANs Through the Lens of Race -- 1 Introduction -- 2 Related Work -- 3 Methodology -- 3.1 Racial Categorizations -- 3.2 Datasets -- 3.3 Amazon Mechanical Turk Annotation -- 4 Experiments and Results -- 4.1 The Racial Distribution of Training and GAN-Generated Data -- 4.2 GAN Quality -- 4.3 Perceived Visual Image Quality and Race -- 5 Discussion -- References -- Trust, but Verify: Using Self-supervised Probing to Improve Trustworthiness -- 1 Introduction -- 2 Related Work -- 2.1 Trustworthiness in Deep Learning -- 2.2 Self-supervised Learning -- 2.3 Probing in Neural Networks -- 3 Methodology -- 3.1 Problem Formulation -- 3.2 Self-supervised Probing Framework -- 4 Experiments -- 4.1 Experimental Setup. ; 4.2 Results on Misclassification Detection.
• Titolo autorizzato: Computer Vision – ECCV 2022
• ISBN: 3-031-19778-X
• Formato: Materiale a stampa
• Livello bibliografico: Monografia
• Lingua di pubblicazione: Inglese
• Record Nr.: 9910629287403321
• Lo trovi qui: Univ. Federico II
• Serie: Lecture notes in computer science.