VISOR Dataset

The authors unveiled the EPIC-KITCHENS VISOR: VIdeo Segmentations and Object Relations, a novel dataset featuring pixel annotations and a benchmark suite tailored for segmenting hands and dynamic objects in egocentric video footage. VISOR introduces an innovative annotation pipeline, incorporating AI-powered elements for enhanced scalability and annotation quality. In total, the authors have made publicly available 272,000 manually annotated semantic masks encompassing 257 object classes, along with 9.9 million interpolated dense masks and 67,000 hand-object relations. This comprehensive dataset covers 36 hours of untrimmed video footage spanning 179 sequences.

作者揭幕了EPIC-KITCHENS VISOR数据集：视频分割与物体关系，该数据集包含像素级标注，并针对在自视角视频素材中分割双手和动态物体而定制了基准测试套件。VISOR引入了一种创新的标注流程，融合了人工智能元素以提升可扩展性和标注质量。总计，作者公开了272,000个手动标注的语义掩码，涵盖257个物体类别，以及990万个插值密集掩码和67,000个手-物体关系。该数据集覆盖了长达36小时的未剪辑视频素材，共计179个序列。

搜集汇总

背景与挑战

背景概述

VISOR数据集是一个针对第一人称视角视频的手部和动态物体语义分割的标注数据集，包含50,729张图像和367,002个标注对象，涵盖305个类别。数据集基于EPIC-KITCHENS-100，提供了丰富的像素级标注和物体关系信息，适用于语义分割、物体检测和识别任务。