Computer Vision – ECCV 2022: 17th European Conference, Tel Aviv, Israel, October 23–27, 2022, Proceedings, Part XXII(13682 Lecture Notes in Computer Science)

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.

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