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

Learning Depth from Focus in the Wild.- Learning-Based Point Cloud Registration for 6D Object Pose Estimation in the Real World.- An End-to-End Transformer Model for Crowd Localization.- Few-Shot Single-View 3D Reconstruction with Memory Prior Contrastive Network.- DID-M3D: Decoupling Instance Depth for Monocular 3D Object Detection.- Adaptive Co-Teaching for Unsupervised Monocular Depth Estimation.- Fusing Local Similarities for Retrieval-Based 3D Orientation Estimation of Unseen Objects.- Lidar Point Cloud Guided Monocular 3D Object Detection.- Structural Causal 3D Reconstruction.- 3D Human Pose Estimation Using M?bius Graph Convolutional Networks.- Learning to Train a Point Cloud Reconstruction Network without Matching.- PanoFormer: Panorama Transformer for Indoor 360° Depth Estimation.- Self-supervised Human Mesh Recovery with Cross-Representation Alignment.- AlignSDF: Pose-Aligned Signed Distance Fields for Hand-Object Reconstruction.- A Reliable Online Method for Joint Estimation of Focal Length and Camera Rotation.- PS-NeRF: Neural Inverse Rendering for Multi-View Photometric Stereo.- Share with Thy Neighbors: Single-View Reconstruction by Cross-Instance Consistency.- Towards Comprehensive Representation Enhancement in Semantics- Guided Self-Supervised Monocular Depth Estimation.- AvatarCap: Animatable Avatar Conditioned Monocular Human Volumetric Capture.- Cross-Attention of Disentangled Modalities for 3D Human Mesh Recovery with Transformers.- GeoRefine: Self-Supervised Online Depth Refinement for Accurate Dense Mapping.- Multi-modal Masked Pre-training for Monocular Panoramic Depth Completion.- GitNet: Geometric Prior-Based Transformation for Birds-Eye View Segmentation.- Learning Visibility for Robust Dense Human Body Estimation.- Towards High-Fidelity Single-View Holistic Reconstructionof Indoor Scenes.- CompNVS: Novel View Synthesis with Scene Completion.- SketchSampler: Sketch-Based 3D Reconstruction via View-Dependent Depth Sampling.- LocalBins: Improving Depth Estimation by Learning Local Distributions.- 2D GANs Meet Unsupervised Single-View 3D Reconstruction.- InfiniteNature-Zero: Learning Perpetual View Generation of Natural Scenes from Single Images.- Semi-Supervised Single-View 3D Reconstruction via Prototype Shape Priors.- Bilateral Normal Integration.- S2Contact: Graph-Based Network for 3D Hand-Object Contact Estimation with Semi-Supervised Learning.- SC-wLS: Towards Interpretable Feed-Forward Camera Re-localization.- FloatingFusion: Depth from ToF and Image-Stabilized Stereo Cameras.- DELTAR: Depth Estimation from a Light-Weight ToF Sensor and RGB Image.- 3D Room Layout Estimation from a Cubemap of Panorama Image via Deep Manhattan Hough Transform.- RBP-Pose: ResidualBounding Box Projection for Category-Level Pose Estimation.- Monocular 3D Object Reconstruction with GAN Inversion.- Map-Free Visual Relocalization: Metric Pose Relative to a Single Image.- Self-Distilled Feature Aggregation for Self-Supervised Monocular Depth Estimation.- Planes vs. Chairs: Category-Guided 3D Shape Learning without Any 3D Cues.