Hand Pose Recognition

[Joshua Blake]

Attached is an image showing my progress on hand pose recognition.

This uses my implementation of the Kinect Body Part Recognition MSR paper. It runs faster than real time with less than one CPU core.

The decision forest used to recognize the hands in this image was trained with 625 computer generated and labeled images representing the combination of five images for each of four degrees of freedom for the hand/wrist range of motion. Training three trees in the forest  took just over 11 hours on a Core i7 2600. Additional degrees of freedom are planned, but take longer to train. I am also looking at post processing the classified pixels to reduce joint jitter.

Thanks,
Josh

Technical Director,
InfoStrat Advanced Technology Group

[RoHS]

Looks great, do you have an executable or any sample code that you
might be willing to share?
-Chris Rojas

>  InfoStratHandPoseRecognition.jpg
> 184KViewDownload

[Joshua Blake]

Hi Chris,

 

I don't have an executable I can share at this point, but I'm working towards that. I need to solve a few more challenges in processing the data and making it more useful. I'll keep you posted as I make more progress and of course when I get the go-ahead to release an executable.

 

By the way, this is written in C# and works with the Kinect SDK or OpenNI.

 

Thanks,

Josh

[Joshua Blake]

Thanks for all the comments so far.

 

I've uploaded a video of this work-in-progress:

http://www.youtube.com/watch?v=Nbc1x76L248

 

Still working on training larger sets of training images and increasing the number of skeleton joints reconstructed.

 

Thanks,

Josh
---
Joshua Blake
Microsoft Surface MVP
OpenKinect Community Founder http://openkinect.org

(cell) 703-946-7176
Twitter: http://twitter.com/joshblake
Blog: http://nui.joshland.org
Natural User Interfaces in .NET book: http://bit.ly/NUIbook