Bonn Activity Maps

Bonn Activity Maps是由波恩大学创建的大规模数据集，专注于多人在厨房场景中的跟踪、活动识别和行为预测。该数据集包含四个不同厨房场景的记录，通过时间同步的摄像头捕捉，重建的3D模型带有语义标注，以及个体人员的运动轨迹和3D人体姿态。数据集通过生成名为活动地图的3D模型上的活动可能性，为研究提供了丰富的资源。其应用领域包括计算机视觉、图形学和机器人学，旨在解决复杂场景分析和理解的问题。

Bonn Activity Maps is a large-scale dataset created by the University of Bonn, focusing on multi-person tracking, activity recognition and behavior prediction in kitchen scenarios. This dataset includes recordings of four distinct kitchen scenes, captured via time-synchronized cameras, alongside semantically annotated reconstructed 3D models, individual human motion trajectories and 3D human poses. It provides abundant research resources by generating activity likelihood maps on 3D models termed activity maps. Its application domains cover computer vision, graphics and robotics, aiming to solve the problems of complex scene analysis and understanding.