UAV-Human
收藏资源简介:
该数据集是由飞行无人机收集的大规模人类行为理解数据集,包含了用于不同任务的各种视频序列和图像。数据集涵盖了多种数据模式,如RGB视频、深度视频、红外序列、骨架数据、鱼眼视频以及夜视视频等。具体规模包括22,476个视频序列,用于姿态估计的22,476张图像,用于重新识别的41,290张图像和1,144个身份信息,以及用于属性识别的22,263张图像。该数据集的任务涵盖了人类动作识别、人体姿态估计、人物重新识别以及人体属性识别等多个领域。
This large-scale human behavior understanding dataset is collected via aerial drones, encompassing diverse video sequences and images for various downstream tasks. The dataset covers multiple data modalities, including RGB videos, depth videos, infrared sequences, skeleton data, fisheye videos, and night-vision videos. In terms of scale, it contains 22,476 video sequences, 22,476 images for human pose estimation, 41,290 images and 1,144 identity annotations for person re-identification, as well as 22,263 images for attribute recognition. The supported tasks span multiple research fields, namely human action recognition, human pose estimation, person re-identification, and human attribute recognition.




