The best Kinect hacks we’ve been seeing are those who are dedicated in making life easier and more convenient especially to those who need them. The Kinect Hand Gesture Recognition using the Neural Network is one the latest hacks we’ve seen which utilizes the depth camera in order to translate and convey hand sign languages. This video by hwassner shows how, with the Neural Network, hand signs can be recognized and even enrolled to create a comprehensive language software. In the video, the sign language counterpart of the words “hello” and “sorry” are now detected by the Kinect. Hand movements that proclaim these words are recognized and are shown to the user. This will definitely bridge the language barrier that has constantly plagued humanity.
“We using the openNI and the NITE librairies with a Kinect to track body parts (hands, elbow, head, …). Then we use FANN (Fast Artificial Neural Network) library for the recognition system.
For now, It is able to recognize only two moves, one means ‘hello’ (“bonjour” in franch), the other mean ‘sorry’ (“désole” in french).
We just need to record new moves to train the neural network to be able to recognize other “words”/moves.”
For more information about the Kinect Hand Gesture Recongition with Neural Network, visit the project’s Youtube Page.