AG-VPReID
收藏资源简介:
AG-VPReID数据集是由昆士兰科技大学等机构创建的大型基准数据集,包含6632个身份标识,32321个轨迹,以及约960万帧图像。该数据集通过无人机、CCTV和可穿戴相机在不同时间和不同视角下捕获,旨在为视频基础上的人员重识别提供具有挑战性的大规模实验数据。数据集的特点包括:拥有最多的身份标识和轨迹数量;融合了航空、地面和可穿戴设备多种平台;具有最高的飞行拍摄高度;提供了丰富的户外场景和真实世界挑战;并具备全面的身份属性信息。
AG-VPReID is a large-scale benchmark dataset developed by institutions including Queensland University of Technology and other relevant organizations. It contains 6632 identities, 32,321 trajectories, and approximately 9.6 million frame images. Captured via drones, closed-circuit television (CCTV) and wearable cameras across different time periods and viewing angles, this dataset aims to provide challenging large-scale experimental data for video-based person re-identification. The key characteristics of this dataset are as follows: it has the largest number of identities and trajectories; it integrates multiple capture platforms including aerial, ground and wearable devices; it features the highest aerial shooting altitude; it provides rich outdoor scenarios and real-world challenges; and it offers comprehensive identity attribute information.




