Computer Vision - Eccv 2022: 17th European Conference, Tel Aviv, Israel, October 23-27, 2022, Proceedings, Part XII

ISBN-13: 9783031197741 / Angielski / Miękka / 2022 / 757 str.

Shai Avidan; Gabriel Brostow; Moustapha Cissé

Opis

The 39-volume set, comprising the LNCS books 13661 until 13699, constitutes the refereed proceedings of the 17th European Conference on Computer Vision, ECCV 2022, held in Tel Aviv, Israel, during October 23–27, 2022. The 1645 papers presented in these proceedings were carefully reviewed and selected from a total of 5804 submissions. The papers deal with topics such as computer vision; machine learning; deep neural networks; reinforcement learning; object recognition; image classification; image processing; object detection; semantic segmentation; human pose estimation; 3d reconstruction; stereo vision; computational photography; neural networks; image coding; image reconstruction; object recognition; motion estimation.

Spis treści

Explicit Model Size Control and Relaxation via Smooth Regularization for Mixed-Precision Quantization.- BASQ: Branch-Wise Activation-Clipping Search Quantization for Sub-4-Bit Neural Networks.- You Already Have It: A Generator-Free Low-Precision DNN Training

Framework Using Stochastic Rounding.- Real Spike: Learning Real-Valued Spikes for Spiking Neural Networks.- FedLTN: Federated Learning for Sparse and Personalized Lottery Ticket Networks.- Theoretical Understanding of the Information Flow on Continual

Learning Performance.- Exploring Lottery Ticket Hypothesis in Spiking Neural Networks.- On the Angular Update and Hyperparameter Tuning of a Scale-Invariant Network.- LANA: Latency Aware Network Acceleration.- RDO-Q: Extremely Fine-Grained Channel-Wise Quantization via Rate-Distortion Optimization.- U-Boost NAS: Utilization-Boosted Differentiable Neural Architecture Search.- PTQ4ViT: Post-Training Quantization for Vision Transformers with Twin Uniform Quantization.- Bitwidth-Adaptive Quantization-Aware Neural Network Training: A Meta-Learning Approach.- Understanding the Dynamics of DNNs Using Graph Modularity.- Latent Discriminant Deterministic Uncertainty.- Making Heads or Tails: Towards Semantically Consistent Visual Counterfactuals.- HIVE: Evaluating the Human Interpretability of Visual Explanations.- BayesCap: Bayesian Identity Cap for Calibrated Uncertainty in Frozen Neural Networks.- SESS: Saliency Enhancing with Scaling and Sliding.- No Token Left Behind: Explainability-Aided Image Classification and Generation.- Interpretable Image Classification with Differentiable Prototypes Assignment.- Contributions of Shape, Texture, and Color in Visual Recognition.- STEEX: Steering Counterfactual Explanations with Semantics.- Are Vision Transformers Robust to Patch Perturbations?.- A Dataset Generation Framework for Evaluating Megapixel Image Classifiers & Their Explanations.- Cartoon Explanations of Image Classifiers.- Shap-CAM: Visual Explanations for Convolutional Neural Networks

Based on Shapley Value.- Privacy-Preserving Face Recognition with Learnable Privacy Budgets in Frequency Domain.- Contrast-Phys: Unsupervised Video-Based Remote Physiological Measurement via Spatiotemporal Contrast.- Source-Free Domain Adaptation with Contrastive Domain Alignment and Self-Supervised Exploration for Face Anti-Spoofing.- On Mitigating Hard Clusters for Face Clustering.- OneFace: One Threshold for All.- Label2Label: A Language Modeling Framework for Multi Attribute Learning.- AgeTransGAN for Facial Age Transformation with Rectified Performance Metrics.- Hierarchical Contrastive Inconsistency Learning for Deepfake Video Detection.- Rethinking Robust Representation Learning under Fine-Grained Noisy Faces.- Teaching Where to Look: Attention Similarity Knowledge Distillation

for Low Resolution Face Recognition.- Teaching with Soft Label Smoothing for Mitigating Noisy Labels in Facial Expressions.- Learning Dynamic Facial Radiance Fields for Few-Shot Talking Head Synthesis.- CoupleFace: Relation Matters for Face Recognition Distillation.- Controllable and Guided Face Synthesis for Unconstrained Face Recognition.- Towards Robust Face Recognition with Comprehensive Search.- Towards Unbiased Label Distribution Learning for Facial Pose Estimation Using Anisotropic Spherical Gaussian.