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

ISBN-13: 9783031198144 / 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

Salient Object Detection for Point Clouds.- Learning Semantic Segmentation from Multiple Datasets with Label Shifts.- Weakly Supervised 3D Scene Segmentation with Region-Level Boundary Awareness and Instance Discrimination.- Towards Open-Vocabulary Scene Graph Generation with Prompt-Based Finetuning.- Variance-Aware Weight Initialization for Point Convolutional Neural Networks.- Break and Make: Interactive Structural Understanding Using LEGO Bricks.- Bi-PointFlowNet: Bidirectional Learning for Point Cloud Based Scene Flow Estimation.- 3DG-STFM: 3D Geometric Guided Student-Teacher Feature Matching.- Video Restoration Framework and Its Meta-Adaptations to Data-Poor Conditions.- MonteBoxFinder: Detecting and Filtering Primitives to Fit a Noisy Point Cloud.- Scene Text Recognition with Permuted Autoregressive Sequence Models.- When Counting Meets HMER: Counting-Aware Network for Handwritten Mathematical Expression Recognition.- Detecting Tampered Scene Text in the Wild.- Optimal Boxes: Boosting End-to-End Scene Text Recognition by Adjusting Annotated Bounding Boxes via Reinforcement Learning.- GLASS: Global to Local Attention for Scene-Text Spotting.- COO: Comic Onomatopoeia Dataset for Recognizing Arbitrary or Truncated Texts.- Language Matters: A Weakly Supervised Vision-Language Pre-training Approach for Scene Text Detection and Spotting.- Toward Understanding WordArt: Corner-Guided Transformer for Scene Text Recognition.- Levenshtein OCR.- Multi-Granularity Prediction for Scene Text Recognition.- Dynamic Low-Resolution Distillation for Cost-Efficient End-to-End Text Spotting.- Contextual Text Block Detection towards Scene Text Understanding.- CoMER: Modeling Coverage for Transformer-Based Handwritten Mathematical Expression Recognition.- Don’t Forget Me: Accurate Background Recovery for Text Removal via Modeling Local-Global Context.- TextAdaIN: Paying Attention to Shortcut Learning in Text Recognizers.- Multi-modal Text Recognition Networks: Interactive Enhancements between Visual and Semantic Features.- SGBANet: Semantic GAN and Balanced Attention Network for Arbitrarily Oriented Scene Text Recognition.- Pure Transformer with Integrated Experts for Scene Text Recognition.- OCR-Free Document Understanding Transformer.- CAR: Class-Aware Regularizations for Semantic Segmentation.- Style-Hallucinated Dual Consistency Learning for Domain Generalized Semantic Segmentation.- SeqFormer: Sequential Transformer for Video Instance Segmentation.- Saliency Hierarchy Modeling via Generative Kernels for Salient Object Detection.- In Defense of Online Models for Video Instance Segmentation.- Active Pointly-Supervised Instance Segmentation.- A Transformer-Based Decoder for Semantic Segmentation with Multi-level Context Mining.- XMem: Long-Term Video Object Segmentation with an Atkinson- Shiffrin Memory Model.- Self-Distillation for Robust LiDAR Semantic Segmentation in Autonomous Driving.- 2DPASS: 2D Priors Assisted Semantic Segmentation on LiDAR Point Clouds.- Extract Free Dense Labels from CLIP.- 3D Compositional Zero-Shot Learning with DeCompositional Consensus.- Video Mask Transfiner for High-Quality Video Instance Segmentation.