Okutama-Action
收藏资源简介:
Okutama-Action数据集是由国家信息学研究所等机构合作创建,专注于从无人机视角捕捉的并发人类动作检测。该数据集包含43个完全标注的视频序列,涵盖12种户外动作类别,总计77365帧,分辨率达3840x2160。数据收集过程中,无人机在不同高度和角度下捕捉视频,确保动作的清晰可见和可区分性。该数据集旨在推动真实世界应用中的动作检测技术,特别是在监控和搜索救援任务中识别和跟踪人类动作的需求。
Okutama-Action Dataset was co-created by the National Institute of Informatics and other collaborating institutions, focusing on concurrent human action detection captured from unmanned aerial vehicle (UAV) perspectives. This dataset contains 43 fully annotated video sequences, covering 12 outdoor action categories, with a total of 77,365 frames and a resolution of 3840×2160. During data collection, drones captured videos at varying altitudes and angles to ensure clear visibility and discriminability of the actions. This dataset aims to advance action detection technologies for real-world applications, particularly catering to the demands of identifying and tracking human actions in surveillance and search-and-rescue tasks.

- 1Okutama-Action: An Aerial View Video Dataset for Concurrent Human Action Detection国家信息学研究所 · 2017年



