ENIGMA-360
收藏资源简介:
ENIGMA-360是由卡塔尼亚大学与Next Vision联合开发的工业场景多视角行为理解数据集,包含180对时空同步的自我中心(ego)与外部视角(exo)视频,总规模达360条。数据采集于真实工业实验室,涵盖高压板组装等典型维护流程,配备精细的时间动作分割、关键步骤识别及人机交互空间标注。该数据集旨在解决工业环境中复杂行为建模的挑战,为智能辅助系统开发提供多视角联合分析基准。
ENIGMA-360 is a multi-view behavior understanding dataset for industrial scenarios, jointly developed by the University of Catania and Next Vision. It contains 180 pairs of spatiotemporally synchronized egocentric and exocentric video clips, with a total of 360 clips. The data is collected in real-world industrial laboratories, covering typical maintenance procedures such as high-voltage board assembly, and is accompanied by fine-grained temporal action segmentation, key step recognition and spatial annotations for human-robot interaction. This dataset aims to address the challenges of complex behavior modeling in industrial environments, providing a multi-view joint analysis benchmark for the development of intelligent assistance systems.



