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

ISBN-13: 9783031200670 / Angielski / Miękka / 2022 / 749 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

UnrealEgo: A New Dataset for Robust Egocentric 3D Human Motion Capture.- Skeleton-Parted Graph Scattering Networks for 3D Human Motion Prediction.- Rethinking Keypoint Representations: Modeling Keypoints and Poses as Objects for Multi-Person Human Pose Estimation.- VirtualPose: Learning Generalizable 3D Human Pose Models from

Virtual Data.- Poseur: Direct Human Pose Regression with Transformers.- SimCC: A Simple Coordinate Classification Perspective for Human Pose Estimation.- Regularizing Vector Embedding in Bottom-Up Human Pose Estimation.- A Visual Navigation Perspective for Category-Level Object Pose Estimation.- Faster VoxelPose: Real-Time 3D Human Pose Estimation by Orthographic Projection.- Learning to Fit Morphable Models.- EgoBody: Human Body Shape and Motion of Interacting People from Head-Mounted Devices.- Grasp’D: Differentiable Contact-Rich Grasp Synthesis for Multi-Fingered Hands.- AutoAvatar: Autoregressive Neural Fields for Dynamic Avatar Modeling.- Deep Radial Embedding for Visual Sequence Learning.- SAGA: Stochastic Whole-Body Grasping with Contact.- Neural Capture of Animatable 3D Human from Monocular Video.- General Object Pose Transformation Network from Unpaired Data.- Compositional Human-Scene Interaction Synthesis with Semantic Control.- PressureVision: Estimating Hand Pressure from a Single RGB Image.- PoseScript: 3D Human Poses from Natural Language.- DProST: Dynamic Projective Spatial Transformer Network for 6D Pose Estimation.- 3D Interacting Hand Pose Estimation by Hand De-Occlusion and Removal.- Pose for Everything: Towards Category-Agnostic Pose Estimation.- PoseGPT: Quantization-Based 3D Human Motion Generation and Forecasting.- DH-AUG: DH Forward Kinematics Model Driven Augmentation for 3D Human Pose Estimation.- Estimating Spatially-Varying Lighting in Urban Scenes with Disentangled Representation.- Boosting Event Stream Super-Resolution with a Recurrent Neural Network.- Projective Parallel Single-Pixel Imaging to Overcome Global Illumination in 3D Structure Light Scanning.- Semantic-Sparse Colorization Network for Deep Exemplar-Based Colorization.- Practical and Scalable Desktop-Based High-Quality Facial Capture.- FAST-VQA: Efficient End-to-End Video Quality Assessment with Fragment Sampling.- Physically-Based Editing of Indoor Scene Lighting from a Single Image.- LEDNet: Joint Low-Light Enhancement and Deblurring in the Dark.- MPIB: An MPI-Based Bokeh Rendering Framework for Realistic Partial Occlusion Effects.- Real-RawVSR: Real-World Raw Video Super-Resolution with a Benchmark Dataset.- Transform Your Smartphone into a DSLR Camera: Learning the ISP in the Wild.- Learning Deep Non-Blind Image Deconvolution without Ground Truths.- NEST: Neural Event Stack for Event-Based Image Enhancement.- Editable Indoor Lighting Estimation.- Fast Two-Step Blind Optical Aberration Correction.- Seeing Far in the Dark with Patterned Flash.- PseudoClick: Interactive Image Segmentation with Click Imitation