RatPose

RatPose是由大连理工大学创建的生物机器人运动预测数据集，专注于通过收集的第三人称视频序列预测生物机器人的运动。该数据集考虑了个体和环境的影响因素，涵盖了5种场景、6个个体和1023对数据。创建过程中，使用头顶摄像头捕捉垂直视角的环境，记录了多种环境下的视频，如迷宫、实验室模拟和开放场地。RatPose数据集旨在解决生物机器人行为控制中的不确定性问题，特别是在长期运动预测和行为控制方面。

RatPose is a bio-robot motion prediction dataset developed by Dalian University of Technology. It focuses on predicting the motion of bio-robots using collected third-person video sequences. This dataset considers the influencing factors of both individual subjects and the environment, covering 5 scenarios, 6 individual bio-robots, and 1023 data pairs. During its construction, overhead cameras were employed to capture top-down perspective environments, and videos were recorded across various settings including mazes, laboratory simulations, and open fields. The RatPose dataset is intended to resolve uncertainty issues in bio-robot behavior control, particularly in the domains of long-term motion prediction and behavior control.