Hand Gesture Recognition using Kinect Sensor

Previously hand gesture recognition was done using glove based techniques. But with the increase in the use of automation, humans preferred recognition methods which did not use any hardware even if it meant that the accuracy was compromised. The Human computer interaction with user is not limited to keyboard, mouse and pen due to evolution of ubiquitous computing. Applications of hand gesture recognition include Phobia therapy, military simulation and medical training. In this project, a two-stage HGR system is proposed to recognize hand using Kinetic sensor which includes segmentation and recognition using both colour and depth information. As most of the mobile cameras used have both depth and normal camera, thus this technique can also be implemented using an app. Here for the recognition part of the project I have used deep learning using autoencoders and verified the algorithm for different training sets