Thanks esuhler )) After two months I need to show the algorithmic basics of the program in a conference in Russia. After that, I think, I can share the code and article with you.
P.S. I still need to improve some parts of the code. Algorithm is really good, but you can still notice some issues, which are cause of the slow speed of recognition.
That would be great! - We are starting to do research work in a similar area.
A few questions -
Are you storing the actual skeletal data arrays from your algorithm as output from the code for analysis later on or is everthing done only in real time?
What process does the user have to go through to register their hands to the Kinect (or any other camera)before it registers the skeletal structure?
Is this using (or have you tried) using the near mode capabilities of different camera solutions?
Now I am working just on skeletal recognition. The final step is to recognize hand gestures via hand skeleton analyses in real time. Surely to do that, patterns of gestures will be stored in DB.
In this demo I am using Kinect for XBox 360, so near mode is not used here.
Months ago, I published one project http://fingertip.codeplex.com/. The code is not written with a good style (results of researche :) ), but I think, it will answer to your other questions.
Thanks - I wil take a look at that. Good luck to you , as well, on finishing the algorithms. I look forward to hearing how your presentation in Russia goes and hearing more after that!
This is excellent. Any chance there is code you would be willing to share?
ReplyDeleteThanks esuhler ))
ReplyDeleteAfter two months I need to show the algorithmic basics of the program in a conference in Russia. After that, I think, I can share the code and article with you.
P.S. I still need to improve some parts of the code. Algorithm is really good, but you can still notice some issues, which are cause of the slow speed of recognition.
Vahagn
That would be great! - We are starting to do research work in a similar area.
DeleteA few questions -
Are you storing the actual skeletal data arrays from your algorithm as output from the code for analysis later on or is everthing done only in real time?
What process does the user have to go through to register their hands to the Kinect (or any other camera)before it registers the skeletal structure?
Is this using (or have you tried) using the near mode capabilities of different camera solutions?
THANKS!!
Now I am working just on skeletal recognition. The final step is to recognize hand gestures via hand skeleton analyses in real time. Surely to do that, patterns of gestures will be stored in DB.
ReplyDeleteIn this demo I am using Kinect for XBox 360, so near mode is not used here.
Months ago, I published one project http://fingertip.codeplex.com/. The code is not written with a good style (results of researche :) ), but I think, it will answer to your other questions.
Good luck in your research!
Vahagn
Thanks - I wil take a look at that. Good luck to you , as well, on finishing the algorithms. I look forward to hearing how your presentation in Russia goes and hearing more after that!
Delete-Ed