Action recognition based on Fast Dynamic-Time warping method

This paper present an approach for recognition of action, based on fast dynamic-time warping method and a feed forward neural network. We use simple to complex approach in action recognition by decomposing to its basic elements. The human body parts motions are tracked and classified individually. The body parts motions are classified using a modified FastDTW, an approximation of DTW that has linear time and space complexity. FastDTW uses a multilevel approach that recursively projects a solution from a coarse resolution and refines the projected solution. These basic motions are used as input in feed forward neural network to recognize the action.