使用Kinect SDK检测身体的任何部分 [英] Any part of the body detection using Kinect SDK

查看:105
本文介绍了使用Kinect SDK检测身体的任何部分的处理方法,对大家解决问题具有一定的参考价值,需要的朋友们下面随着小编来一起学习吧!

问题描述

您好我正在使用SDK 1.8并试图找到一种方法来检测人体的任何部分。  我在这里使用Skeltiondetection逻辑。

Hi I am using SDK 1.8 and trying to find a way to detect any part of the human body.  I am using the Skeltiondetection logic here.

ex:如果我使用Skeltion.TrackingState == SkeltionTrackingState.Tracked,并计算非空框架,如果我面对kinect相机,它工作正常,运动我来到kinect侧面角度,它没有检测到我的身体。  我
已经尝试将trackingmode作为默认值并且已就位。  但没有运气。  任何线索?

for ex: if I use Skeltion.TrackingState == SkeltionTrackingState.Tracked, and count the non empty frames, and if I am facing the kinect camera, it is working fine, the movement I come infront of kinect in a side angle, it is not detecting my body.  I have tried both trackingmode as default and seated.  But no luck.  Any clue?

谢谢,

Harish V。

Harish Vajja

Harish Vajja

推荐答案

调整Kinect骨架跟踪(ST)系统训练数据,以便用户面对传感器。获得最初的"锁定"通常你需要在坐姿模式下上身的人的全景。一旦获得了这一点,从那一点开始的运动
前进应该继续阻止关节的任何干扰/闭塞。

The Kinect skeleton tracking(ST) system training data is tuned for users to face the sensor. To get the initial "lock on" typically you need the full view of the person, in seated mode the upper body. Once that has been acquired, movements from that point forward should continue baring any interference/occlusion of joints.

ST的替代方案是对深度数据进行自己的计算机视觉算法,以绘制点,无论它们与传感器的方向如何。如果默认行为不符合您特定的
用例,可以找到第三方库。

The alternative to ST would be to do your own computer vision algorithm on the depth data to plot points regardless of their orientation to the sensor. There may be third party libraries you can find if the default behavior does not fit your particular use case.


这篇关于使用Kinect SDK检测身体的任何部分的文章就介绍到这了,希望我们推荐的答案对大家有所帮助,也希望大家多多支持IT屋!

查看全文
登录 关闭
扫码关注1秒登录
发送“验证码”获取 | 15天全站免登陆