Multi-user activity recognition database (First part)

First part of the data repository related to the publication "Towards Multi-User Activity Recognition through Facilitated Training Data and Deep Learning for Human-Robot Collaboration Applications". The remaining part of the dataset is present at another repository with the same name on Figshare.The data is composed by .csv files with the datastream from the Azure Kinect camera used in the experiment. The files starting with "s##_a##" have 3 columns each, referring to the spatial coordinates of the body frames. In the file name, "s##" identifies the subject, "a##" the activity (01: working, 02: requesting, 03: preparing), "t##" the task (not used for this work) and "r##" the repetition.If you are interested in the raw rosbag2 files, please contact the developer Francesco Semeraro at [e-mail]

本数据集为论文《面向人机协作应用的基于增强训练数据与深度学习的多用户活动识别》（Towards Multi-User Activity Recognition through Facilitated Training Data and Deep Learning for Human-Robot Collaboration Applications）相关数据仓库的第一部分。该数据集的剩余部分存储于Figshare平台上的同名数据仓库中。本数据集由实验中使用的Azure Kinect相机（Azure Kinect）采集的数据流对应的.csv格式文件组成。文件名以“s##_a##”开头的文件均包含3列数据，分别对应人体骨骼帧的空间坐标。在文件名中，“s##”代表受试者编号，“a##”代表活动类型（01为作业，02为请求，03为准备），“t##”代表任务（本研究未使用该字段），“r##”代表重复次数。若需获取原始rosbag2格式文件，请联系开发者Francesco Semeraro，联系邮箱为[e-mail]