spianmo / GaitStudio

Based on the Kinect depth camera and MediaPipe, real-time reasoning of the joint activity of each part of the human body, and output a detection report.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

GaitStudio

基于Kinect深度相机和MediaPipe实时推理人体各部分关节活动度,并输出PDF检测报告。

Based on the Kinect depth camera and MediaPipe, real-time reasoning of the joint activity of each part of the human body, and output a detection report.

定义了一套检测规则DSL,可扩展多种检测项目。检测速率在15FPS左右,支持深度视图、RGB视图以及叠加视图的切换显示。实时展示各部位关节角度,自由调节检测BlazePose模型精度和Kinect相机参数。程序入口:HealBone_GaitStudio.py

TODO:

  • 根据关节角度Detect步态相位,目前使用空间点分解加速度的方式,通过加速度滤波得出周期性的步态相位,数据来源于CV推理,不比IMU。

技术栈:

Python3, PySide2, MediaPipe, QT

截图:

image

image

image

6a87ea890718ca79c54a0da4ea74730

About

Based on the Kinect depth camera and MediaPipe, real-time reasoning of the joint activity of each part of the human body, and output a detection report.

License:GNU Affero General Public License v3.0


Languages

Language:Jupyter Notebook 68.0%Language:C 11.3%Language:C++ 8.7%Language:Python 5.1%Language:JavaScript 5.0%Language:CSS 0.8%Language:Roff 0.6%Language:HTML 0.3%Language:Makefile 0.1%Language:CMake 0.1%Language:Batchfile 0.0%