Human activity recognition using skeleton data and support vector machine

In this paper, we propose a method for recognizing human activities using skeleton data by RGB-D camera, namely Kinect device. The human activity recognition is a learning in the field of computer vision. In its application, the recognition of human activity can be used for a sign language learning, human-computer interaction, surveillance of the elderly, image processing and etc. Our approach is based on skeleton data with coordinate value of each joints in human body, that will be classified using support vector machine algorithm when performing a movement to predict the activities name. Experiments were performed with a new training data that we've create manual from capturing movement while human target are doing activities. Experiments result show that the system best average accuracy is 93.75% of all activities prediction with the optimal distance of object to the devices is 2 meters.