-
开发环境
- macOS
- python 3.6
- libfreenect
- OpenCV 3.3
- Kinect v1
- 安装
- Tutorial/在virtualenv安装freenect库-python3.md
-
参考图书:
-
校准
-
常问问题
- 深度图像(v1)中有什么阴影?Kinect阴影图
- kinect可以看到的深度范围是多少?(v1)〜0.7-6米或2.3-20英尺。请注意,在距离太远和太近的两个元素上都会得到黑色像素(或2048的原始深度值)。
-
fakenect hands freenect-cppview
-
Skeleton Tracking骨骼追踪
- 需要使用OpenNi和NiTE
- NiViewer2
- 只支持Kinect V2 ??
-
OpenKinect/libfreenect does not provide skeleton tracking.
-
OpenNI provides skeleton tracking through the proprietary NiTE.
-
The Microsoft SDK provides skeleton tracking as well, it basically the same as the skeleton tracking through NiTE.
- 第2个开源方案:https://github.com/joaquimrocha/Skeltrack
-
Be sure to install GFreenect before Skeltrack https://github.com/elima/GFreenect
-
Once Skeltrack is compiled run the kinect binary in the examples folder
- 需要使用OpenNi和NiTE
-
libfreenect multiple kinects同时连接多个Kinect
- 代码参考
- OpenKinect/libfreenect#522
-
It is somewhat hard to tell about your PCIe bandwidth带宽不够. It is part of the motherboard configuration and it depends on the USB controller link too. You can try out different machines though, or you can use multiple machines and connect them through ROS.
- 需要一台配备强大GPU的高端个人电脑,每个Kinect需要一个USB3控制器,每个控制器都有自己的PCI-Express x8或x16连接。
- 亚马孙 4 Port PCI Express (PCIe) SuperSpeed USB 3.0 Card Adapter
- https://forum.openframeworks.cc/t/multiple-kinect-setup-for-real-time-volumetric-reconstruction-and-tracking-of-people/15271
while(1):
img = get_depth()
freenect.stop_depth(device1)
# display the img
# proceed without delay
freenect.start_depth(device1)
- 20180305