A novel and intelligent method for human detection using laser is presented in this paper. In this method the information is elicited from multiple frames instead of single frames which provides little information. Time has been also added to the data as the third dimension so that the data converted to 3-D data and then were used as short time series to detect 3-D objects. Finally, the data are clustered and then classified using a deep learning network which is called stacked auto encoder. The results show a better and improved performance for this method.
