Computer Vision – ECCV 2022: 17th European Conference, Tel Aviv, Israel, October 23–27, 2022, Proceedings, Part II

ISBN-13: 9783031200854 / Angielski / Miękka / 2022 / 751 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

MHR-Net: Multiple-Hypothesis Reconstruction of Non-rigid Shapes from 2D Views.- Depth Map Decomposition for Monocular Depth Estimation.- Monitored Distillation for Positive Congruent Depth Completion.- Resolution-Free Point Cloud Sampling Network with Data Distillation.- Organic Priors in Non-rigid Structure from Motion.- Perspective Flow Aggregation for Data-Limited 6D Object Pose Estimation.- DANBO: Disentangled Articulated Neural Body Representations via Graph Neural Networks.- CHORE: Contact, Human and Object REconstruction from a Single RGB Image.- Learned Vertex Descent: A New Direction for 3D Human Model Fitting.- Self-Calibrating Photometric Stereo by Neural Inverse Rendering.- 3D Clothed Human Reconstruction in the Wild.- Directed Ray Distance Functions for 3D Scene Reconstruction.- Object Level Depth Reconstruction for Category Level 6D Object Pose Estimation from Monocular RGB Image.- Uncertainty Quantification in Depth Estimation via Constrained Ordinal Regression.- CostDCNet: Cost Volume Based Depth Completion for a Single RGB-D Image.- ShAPO: Implicit Representations for Multi-Object Shape, Appearance, and Pose Optimization.- 3D Siamese Transformer Network for Single Object Tracking on Point Clouds.- Object Wake-Up: 3D Object Rigging from a Single Image.- IntegratedPIFu: Integrated Pixel Aligned Implicit Function for Single-View Human Reconstruction.- Realistic One-Shot Mesh-Based Head Avatars.- A Kendall Shape Space Approach to 3D Shape Estimation from 2D Landmarks.- Neural Light Field Estimation for Street Scenes with Differentiable Virtual Object Insertion.- Perspective Phase Angle Model for Polarimetric 3D Reconstruction.- DeepShadow: Neural Shape from Shadow.- Camera Auto-Calibration from the Steiner Conic of the Fundamental Matrix.- Super-Resolution 3D Human Shape from a Single Low-Resolution Image.- Minimal Neural Atlas: Parameterizing Complex Surfaces with Minimal Charts and Distortion.- ExtrudeNet: Unsupervised Inverse Sketch-and-Extrude for Shape Parsing.- CATRE: Iterative Point Clouds Alignment for Category-Level Object Pose Refinement.- Optimization over Disentangled Encoding: Unsupervised Cross-Domain Point Cloud Completion via Occlusion Factor Manipulation.- Unsupervised Learning of 3D Semantic Keypoints with Mutual Reconstruction.- MvDeCor: Multi-View Dense Correspondence Learning for Fine-Grained 3D Segmentation.- SUPR: A Sparse Unified Part-Based Human Representation.- Revisiting Point Cloud Simplification: A Learnable Feature Preserving Approach.- Masked Autoencoders for Point Cloud Self-Supervised Learning.- Intrinsic Neural Fields: Learning Functions on Manifolds.- Skeleton-Free Pose Transfer for Stylized 3D Characters.- Masked Discrimination for Self-Supervised Learning on Point Clouds.- FBNet: Feedback Network for Point Cloud Completion.- Meta-Sampler: Almost-Universal yet Task-Oriented Sampling for Point Clouds.-A Level Set Theory for Neural Implicit Evolution under Explicit Flows.- Efficient Point Cloud Analysis Using Hilbert Curve.