Computer Vision – ECCV 2022

Explicit Model Size Control and Relaxation via Smooth Regularization&nbsp;for Mixed-Precision Quantization.-&nbsp;BASQ: Branch-Wise Activation-Clipping Search Quantization for&nbsp;Sub-4-Bit Neural Networks.-&nbsp;You Already Have It: A Generator-Free Low-Precision DNN Training<div>Framework Using Stochastic Rounding.-&nbsp;Real Spike: Learning Real-Valued Spikes for Spiking Neural Networks.-&nbsp;FedLTN: Federated Learning for Sparse and Personalized Lottery&nbsp;Ticket Networks.-&nbsp;Theoretical Understanding of the Information Flow on Continual</div><div>Learning Performance.-&nbsp;Exploring Lottery Ticket Hypothesis in Spiking Neural Networks.-&nbsp;On the Angular Update and Hyperparameter Tuning of a&nbsp;Scale-Invariant Network.-&nbsp;LANA: Latency Aware Network Acceleration.-&nbsp;RDO-Q: Extremely Fine-Grained Channel-Wise Quantization via&nbsp;Rate-Distortion Optimization.-&nbsp;U-Boost NAS: Utilization-Boosted Differentiable Neural Architecture&nbsp;Search.-&nbsp;PTQ4ViT: Post-Training Quantization for Vision Transformers with&nbsp;Twin Uniform Quantization.-&nbsp;Bitwidth-Adaptive Quantization-Aware Neural Network Training: A&nbsp;Meta-Learning Approach.-&nbsp;Understanding the Dynamics of DNNs Using Graph Modularity.-&nbsp;Latent Discriminant Deterministic Uncertainty.-&nbsp;Making Heads or Tails: Towards Semantically Consistent Visual&nbsp;Counterfactuals.-&nbsp;HIVE: Evaluating the Human Interpretability of Visual Explanations.-&nbsp;BayesCap: Bayesian Identity Cap for Calibrated Uncertainty in Frozen&nbsp;Neural Networks.-&nbsp;SESS: Saliency Enhancing with Scaling and Sliding.-&nbsp;No Token Left Behind: Explainability-Aided Image Classification and&nbsp;Generation.-&nbsp;Interpretable Image Classification with Differentiable Prototypes&nbsp;Assignment.-&nbsp;Contributions of Shape, Texture, and Color in Visual Recognition.-&nbsp;STEEX: Steering Counterfactual Explanations with Semantics.-&nbsp;Are Vision Transformers Robust to Patch Perturbations?.-&nbsp;A Dataset Generation Framework for Evaluating Megapixel Image&nbsp;Classifiers & Their Explanations.-&nbsp;Cartoon Explanations of Image Classifiers.-&nbsp;Shap-CAM: Visual Explanations for Convolutional Neural Networks</div><div>Based on Shapley Value.-&nbsp;Privacy-Preserving Face Recognition with Learnable Privacy Budgets&nbsp;in Frequency Domain.-&nbsp;Contrast-Phys: Unsupervised Video-Based Remote Physiological&nbsp;Measurement via Spatiotemporal Contrast.-&nbsp;Source-Free Domain Adaptation with Contrastive Domain Alignment&nbsp;and Self-Supervised Exploration for Face Anti-Spoofing.-&nbsp;On Mitigating Hard Clusters for Face Clustering.-&nbsp;OneFace: One Threshold for All.-&nbsp;Label2Label: A Language Modeling Framework for Multi Attribute&nbsp;Learning.-&nbsp;AgeTransGAN for Facial Age Transformation with Rectified&nbsp;Performance Metrics.-&nbsp;Hierarchical Contrastive Inconsistency Learning for Deepfake Video&nbsp;Detection.-&nbsp;Rethinking Robust Representation Learning under Fine-Grained Noisy&nbsp;Faces.-&nbsp;Teaching Where to Look: Attention Similarity Knowledge Distillation</div><div>for Low Resolution Face Recognition.-&nbsp;Teaching with Soft Label Smoothing for Mitigating Noisy Labels in&nbsp;Facial Expressions.-&nbsp;Learning Dynamic Facial Radiance Fields for Few-Shot Talking Head&nbsp;Synthesis.-&nbsp;CoupleFace: Relation Matters for Face Recognition Distillation.-&nbsp;Controllable and Guided Face Synthesis for Unconstrained Face Recognition.-&nbsp;Towards Robust Face Recognition with Comprehensive Search.-&nbsp;Towards Unbiased Label Distribution Learning for Facial Pose&nbsp;Estimation Using Anisotropic Spherical Gaussian.</div>