Moving Object Tracking Using Machine Learning
Author : Seethal Prince E, Asha S and Greshma P Sebastian
Abstract :
Object tracking is the process of locating the object which is in motion. It is an important task in the field of computer vision. Due to the variations in pose, size, illumination, motion of object and the like, object tracking becomes a challenging mission. Widely established tracking method is tracking-by- detection method. Object may change its appearance throughout the image sequences or video. If the tracker is not learning this change in the appearance dynamically, the chance for drift is high and that may leads to reduce the tracker efficiency. To overcome this issue, this study proposing a method to adopt the change in appearance of a moving object with Active Appearance Model. To represent the object in more natural manner SVM based Multiple Instance Learning (MIL) representation is also used. The location of the moving object in the first frame is given as the input to the tracker. The proposed method achieves good result with real time performance. The error plots and the precision plots for different standard object tracking datasets are generated.
Keywords :
Object tacking, detection method, multiple instance learning (MIL)
