Expi

该数据集包含了115个序列，记录了两对专业舞者表演16种不同舞蹈动作的过程，以每秒25帧的速度进行拍摄。此外，该数据集在常见动作和未见动作的分割上都进行了基准测试。任务是对多姿态动作进行预测。

This dataset contains 115 sequences capturing the performance of 16 distinct dance movements by two pairs of professional dancers, filmed at 25 frames per second. Additionally, this dataset has been benchmarked on action segmentation tasks for both seen and unseen actions, with the core task being multi-pose action prediction.