Visual tracking is the way towards finding, recognizing and deciding the dynamic design of one or numerous moving(potentially deforma-ble) protests (or parts of items) in each casing of one or a few cameras. To build a general tracking system, the recent progress in represen-tation of image, shape, object, target, appearance model and motion model are briefly reviewed in this paper. For following the single target or the various targets the models that is inspected here is basic enough to be applied. The appearance model which is trending in recent time given a special attention. The key techniques and the factors which is tedious for the tracker to find the object appearance changes are camera motion, illumination variation, shape deformation and partial occlusion is discussed. For tracking -by-detection and on-line boosting methods we see that the state-of-the-art performs well (e. g TLD, Online Boost, MIL-Track,). Hence, based on this we check them together for a single person tracking.
Bhuvaneshwari et al. (Fri,) studied this question.