Computer Vision – ECCV 2022

ByteTrack: Multi-Object Tracking by Associating Every Detection Box.-&nbsp;Robust Multi-Object Tracking by Marginal Inference.-&nbsp;PolarMOT: How Far Can Geometric Relations Take Us in 3D&nbsp;Multi-Object Tracking?.-&nbsp;Particle Video Revisited: Tracking through Occlusions Using Point&nbsp;Trajectories.-&nbsp;Tracking Objects As Pixel-Wise Distributions.-&nbsp;CMT: Context-Matching-Guided Transformer for 3D Tracking in Point&nbsp;Clouds.-&nbsp;Towards Generic 3D Tracking in RGBD Videos: Benchmark and Baseline.-&nbsp;Hierarchical Latent Structure for Multi-modal Vehicle Trajectory&nbsp;Forecasting.-&nbsp;AiATrack: Attention in Attention for Transformer Visual Tracking.-&nbsp;Disentangling Architecture and Training for Optical Flow.-&nbsp;A Perturbation-Constrained Adversarial Attack for Evaluating the&nbsp;Robustness of Optical Flow.-&nbsp;Robust Landmark-Based Stent Tracking in X-Ray Fluoroscopy.-&nbsp;Social ODE: Multi-agent Trajectory Forecasting with Neural Ordinary&nbsp;Differential Equations.-&nbsp;Social-SSL: Self-Supervised Cross-Sequence Representation Learning&nbsp;Based on Transformers for Multi-agent Trajectory Prediction.-&nbsp;Diverse Human Motion Prediction Guided by Multi-level Spatial-&nbsp;Temporal Anchors.-&nbsp;Learning Pedestrian Group Representations for Multi-modal Trajectory&nbsp;Prediction.-&nbsp;Sequential Multi-View Fusion Network for Fast LiDAR Point Motion&nbsp;Estimation.-&nbsp;E-Graph: Minimal Solution for Rigid Rotation with Extensibility Graphs.-&nbsp;Point Cloud Compression with Range Image-Based Entropy Model for<div>Autonomous Driving.-&nbsp;Joint Feature Learning and Relation Modeling for Tracking: A&nbsp;One-Stream Framework.-&nbsp;MotionCLIP: Exposing Human Motion Generation to CLIP Space.-&nbsp;Backbone Is All Your Need: A Simplified Architecture for Visual&nbsp;Object Tracking.-&nbsp;Aware of the History: Trajectory Forecasting with the Local Behavior&nbsp;Data.-&nbsp;Optical Flow Training under Limited Label Budget via Active Learning.-&nbsp;Hierarchical Feature Embedding for Visual Tracking.-&nbsp;Tackling Background Distraction in Video Object Segmentation.-&nbsp;Social-Implicit: Rethinking Trajectory Prediction Evaluation and the&nbsp;Effectiveness of Implicit Maximum Likelihood Estimation.-&nbsp;TEMOS: Generating Diverse Human Motions from Textual Descriptions.-&nbsp;Tracking Every Thing in the Wild.-&nbsp;HULC: 3D HUman Motion Capture with Pose Manifold SampLing and&nbsp;Dense Contact Guidance.-&nbsp;Towards Sequence-Level Training for Visual Tracking.-&nbsp;Learned Monocular Depth Priors in Visual-Inertial Initialization.-&nbsp;Robust Visual Tracking by Segmentation.-&nbsp;MeshLoc: Mesh-Based Visual Localization.-&nbsp;S2F2: Single-Stage Flow Forecasting for Future Multiple Trajectories&nbsp;Prediction.-&nbsp;Large-Displacement 3D Object Tracking with Hybrid Non-local&nbsp;Optimization.-&nbsp;FEAR: Fast, Efficient, Accurate and Robust Visual Tracker.-&nbsp;PREF: Predictability Regularized Neural Motion Fields.-&nbsp;View Vertically: A Hierarchical Network for Trajectory Prediction via&nbsp;Fourier Spectrums.-&nbsp;HVC-Net: Unifying Homography, Visibility, and Confidence Learning&nbsp;for Planar Object Tracking.-&nbsp;RamGAN: Region Attentive Morphing GAN for Region-Level Makeup&nbsp;Transfer.-&nbsp;SinNeRF: Training Neural Radiance Fields on Complex Scenes from a&nbsp;Single Image.-&nbsp;Entropy-Driven Sampling and Training Scheme for Conditional&nbsp;Diffusion Generation.</div>