Because Kinect is still alive we decided to share our KinectV2 library.
Speed comes first, so we tried to create code that let’s us choose what features we would need for a project (color vs no color, only 2d vs 3d tracking, etc…). It includes stuff like user management, 2d and 3d scene management, sending data over the network for multiple kinects, pointcloud triangulation, gesture detection, area management, etc…
Wow looks great!
How hard is it to train gestures?
We used the Visual Gesture Builder. It comes with the KinectV2 sdk.
We used a lot of trial and error because it is undocumented and Microsoft has never taken it into production phase and is not supporting it anymore. But it still works.
Another library we sometimes use is the very good Gesture Recognition Toolkit by Nick Gillian.
Look for DTW (dynamic time warping)
Great! thanks for the info
hey @wimvanhenden ,
thanks a lot for sharing, it’s working perfect.
it’s really straight forward to use.