Computer Vision – ECCV 2022

Expanding Language-Image Pretrained Models for General Video&nbsp;Recognition.-&nbsp;Hunting Group Clues with Transformers for Social Group Activity&nbsp;Recognition.-&nbsp;Contrastive Positive Mining for Unsupervised 3D Action Representation&nbsp;Learning.-&nbsp;Target-Absent Human Attention.-&nbsp;Uncertainty-Based Spatial-Temporal Attention for Online Action Detection.-&nbsp;Iwin: Human-Object Interaction Detection via Transformer with Irregular Windows.-&nbsp;Rethinking Zero-Shot Action Recognition: Learning from Latent Atomic Actions.-&nbsp;Mining Cross-Person Cues for Body-Part Interactiveness Learning in HOI Detection.-&nbsp;Collaborating Domain-Shared and Target-Specific Feature Clustering for Cross-Domain 3D Action Recognition.-&nbsp;Is Appearance Free Action Recognition Possible?.-&nbsp;Learning Spatial-Preserved Skeleton Representations for Few-Shot<div>Action Recognition.-&nbsp;Dual-Evidential Learning for Weakly-Supervised Temporal Action</div><div>Localization.-&nbsp;Global-Local Motion Transformer for Unsupervised Skeleton-Based</div><div>Action Learning.-&nbsp;AdaFocusV3: On Unified Spatial-Temporal Dynamic Video Recognition.-&nbsp;Panoramic Human Activity Recognition.-&nbsp;Delving into Details: Synopsis-to-Detail Networks for Video Recognition.-&nbsp;A Generalized & Robust Framework for Timestamp Supervision in&nbsp;Temporal Action Segmentation.-&nbsp;Few-Shot Action Recognition with Hierarchical Matching and&nbsp;Contrastive Learning.-&nbsp;PrivHAR: Recognizing Human Actions from Privacy-Preserving Lens.-&nbsp;Scale-Aware Spatio-Temporal Relation Learning for Video Anomaly&nbsp;Detection.-&nbsp;Compound Prototype Matching for Few-Shot Action Recognition.-&nbsp;Continual 3D Convolutional Neural Networks for Real-Time Processing&nbsp;of Videos.-&nbsp;Dynamic Spatio-Temporal Specialization Learning for Fine-Grained&nbsp;Action Recognition.-&nbsp;Dynamic Local Aggregation Network with Adaptive Clusterer for Anomaly Detection.-&nbsp;Action Quality Assessment with Temporal Parsing Transformer.-&nbsp;Entry-Flipped Transformer for Inference and Prediction of Participant&nbsp;Behavior.-&nbsp;Pairwise Contrastive Learning Network for Action Quality Assessment.-&nbsp;Geometric Features Informed Multi-Person Human-Object Interaction&nbsp;Recognition in Videos.-&nbsp;ActionFormer: Localizing Moments of Actions with Transformers.-&nbsp;SocialVAE: Human Trajectory Prediction Using Timewise Latents.-&nbsp;Shape Matters: Deformable Patch Attack.-&nbsp;Frequency Domain Model Augmentation for Adversarial Attack.-&nbsp;Prior-Guided Adversarial Initialization for Fast Adversarial Training.-&nbsp;Enhanced Accuracy and Robustness via Multi-Teacher Adversarial Distillation.-&nbsp;LGV: Boosting Adversarial Example Transferability from Large&nbsp;Geometric Vicinity.-&nbsp;A Large-Scale Multiple-Objective Method for Black-Box Attack against Object Detection.-&nbsp;GradAuto: Energy-Oriented Attack on Dynamic Neural Networks.-&nbsp;A Spectral View of Randomized Smoothing under Common&nbsp;Corruptions: Benchmarking and Improving Certified Robustness.-&nbsp;Improving Adversarial Robustness of 3D Point Cloud Classification Models.-&nbsp;Learning Extremely Lightweight and Robust Model with Differentiable&nbsp;Constraints on Sparsity and Condition Number.-&nbsp;RIBAC: Towards Robust and Imperceptible Backdoor Attack against&nbsp;Compact DNN.-&nbsp;Boosting Transferability of Targeted Adversarial Examples via&nbsp;Hierarchical Generative Networks.</div>