遇见数据集

Vogtareuth Rehabilitation Pose Dataset

收藏
Zenodo2025-10-21 更新2026-05-26 收录
官方服务:

资源简介:

Vogtareuth Rehablitation (VtR) is a large-scale human posture dataset specific to physical rehablitation. It contains depth images and 2D and 3D pose annotations. The dataset has varied samples of children and a few adults, different genders, a variety of poses such as standing, sitting, bending, etc. and also instances of limited range of motion for some joints. 12 children (6 males and 6 females) were recorded using an Azure Kinect sensor while they played games designed and developed by reFit for rehabilitation. The actions involved in the games are sideways trunk movement in the frontal plane, i.e., torso abduction and adduction, arm abduction and adduction, arm flexion and extension, torso rotation, knee raise, bending, and some free movements. The pose annotations were extracted using the Azure Kinect Body Tracking SDK . The 3D poses available in this dataset are in the camera coordinate space and the 2D poses are in the depth image space. Altogether 25 joints were recorded. Skeleton Structure of VtR Dataset SPINE_BASE 1 SPINE_MID 2 NECK 3 HEAD 4 SHOULDER_LEFT 5 ELBOW_LEFT 6 WRIST_LEFT 7 NOSE 8 SHOULDER_LEFT 9 ELBOW_RIGHT 10 WRIST_RIGHT 11 HAND_RIGHT 12 HIP_LEFT 13 KNEE_LEFT 14 ANKLE_LEFT 15 FOOT_LEFT 16 HIP_RIGHT 17 KNEE_RIGHT 18 ANKLE_RIGHT 19 FOOT_RIGHT 20 SPINE_SHOULDER 21 HANDTIP_LEFT 22 THUMB_LEFT 23 HANDTIP_RIGHT 24 THUMB_RIGHT 25 30 recordings are provided in this dataset. Every recording has three associated files with the following naming convention. Record_<Recording_Date>_<UserId>_<Recording_Number>_depth.zip: contains the anonymized depth images. Record_<Recording_Date>_<UserId>_<Recording_Number>_export_2D_depth.csv: contains the 2D posture in the depth image space. Record_<Recording_Date>_<UserId>_<Recording_Number>_export_3D.csv: contains the 3D posture in camera coordinate space.

沃格托罗伊特康复(VtR)数据集是一款专为物理康复场景打造的大规模人体姿态数据集,涵盖深度图像以及二维、三维姿态标注。该数据集包含多样化的样本:包含儿童与少量成人受试者,覆盖不同性别,囊括站立、坐姿、弯腰等多种姿态类型,同时包含部分关节活动范围受限的样本案例。 研究团队使用Azure Kinect传感器,对12名儿童(6名男性、6名女性)进行数据采集,受试者在采集过程中参与由reFit开发的康复游戏。游戏包含的动作包括:额状面内躯干侧移(即躯干外展与内收)、手臂外展与内收、手臂屈伸、躯干旋转、提膝、弯腰,以及若干自由动作。 姿态标注通过Azure Kinect Body Tracking SDK提取完成。本数据集提供的三维姿态数据处于相机坐标系空间,二维姿态数据则处于深度图像坐标系空间。本次采集共记录了25个关节点。 ### VtR数据集骨骼结构 1. 脊柱基底(SPINE_BASE) 2. 脊柱中段(SPINE_MID) 3. 颈部(NECK) 4. 头部(HEAD) 5. 左肩(SHOULDER_LEFT) 6. 左肘(ELBOW_LEFT) 7. 左手腕(WRIST_LEFT) 8. 鼻尖(NOSE) 9. 左肩(SHOULDER_LEFT,原文疑似笔误,应为右肩) 10. 右肘(ELBOW_RIGHT) 11. 右手腕(WRIST_RIGHT) 12. 右手(HAND_RIGHT) 13. 左髋(HIP_LEFT) 14. 左膝(KNEE_LEFT) 15. 左脚踝(ANKLE_LEFT) 16. 左脚(FOOT_LEFT) 17. 右髋(HIP_RIGHT) 18. 右膝(KNEE_RIGHT) 19. 右脚踝(ANKLE_RIGHT) 20. 右脚(FOOT_RIGHT) 21. 肩脊柱(SPINE_SHOULDER) 22. 左手指尖(HANDTIP_LEFT) 23. 左手拇指(THUMB_LEFT) 24. 右手指尖(HANDTIP_RIGHT) 25. 右手拇指(THUMB_RIGHT) 本数据集共提供30组采集记录。每组记录包含三个关联文件,命名规则如下: - `Record_<录制日期>_<用户ID>_<录制编号>_depth.zip`:包含已匿名化的深度图像。 - `Record_<录制日期>_<用户ID>_<录制编号>_export_2D_depth.csv`:包含深度图像坐标系下的二维姿态数据。 - `Record_<录制日期>_<用户ID>_<录制编号>_export_3D.csv`:包含相机坐标系下的三维姿态数据。

提供机构:
Zenodo
创建时间:
2025-10-21
二维码
社区交流群
二维码
科研交流群
商业服务