Complex and Long Activities Dataset (CLAD)

Complex and Long Activities Dataset (CLAD) 是由利兹大学和Five AI创建的一个包含62个视频的数据集，旨在研究复杂和长期的人类活动。数据集通过Kinect 2传感器记录，包含RGB-D图像、点云数据和自动生成的骨骼跟踪，以及通过众包获得的丰富注释。CLAD特别适合作为活动识别研究的测试平台，也可应用于机器人/计算机视觉研究中的其他任务，如对象检测和人类骨骼跟踪。数据集的创建过程涉及自然和非脚本的活动记录，以及通过众包进行注释的有效机制。其应用领域包括长期自主机器人系统的活动理解和复杂活动分析。

Complex and Long Activities Dataset (CLAD) was developed by the University of Leeds and Five AI, consisting of 62 videos and tailored for research on complex and long-term human activities. Recorded using Kinect 2 sensors, the dataset includes RGB-D images, point cloud data, automatically generated skeletal tracking results, as well as rich annotations obtained via crowdsourcing. CLAD is particularly suitable as a testbed for activity recognition research, and can also be applied to other tasks in robotics and computer vision research, such as object detection and human skeletal tracking. The dataset creation process involves recording natural, unscripted activities and an effective crowdsourcing-based annotation mechanism. Its application areas include activity understanding for long-term autonomous robotic systems and complex activity analysis.