WPF(Windows Presentation Foundation) application for Windows using Kinect sensor and Kinect SDK. Implemented with Google Translator to get the actual voice representation of gestures.
This is a system for training sign language gestures and enable comupter systems experimentaly evaluate results of recognition and determine which gesture of language it is. The system is based on Microsoft Kinect sensor and dynamic time warping algorithm.
The system provides some advantages. By using Kinect sensor there is no need to wear any form of an armband or sleeve. The algorithm itself can be used on all sequences that can be vectorized. It can also align two sequences that are not aligned so recognition is more robust.
There are also some disadvantages to consider. With growing number of gestures, the accuracy of recognition slightly decreases. Therefore the system is efective for small portion of gestures. This disadvantages might be improved by using virtual reality system which detects fingers position.