Integrate Occlusion Control Block to Your Tracker

Visual object tracking is a hot topic in computer vision, focusing on localization of target object and size of a target object across successive frames. However, this endeavor faces a couple of challenges, including occlusion, illumination, scale variations, background clutter, and detecting small objects. In these challenges, we focus on occlusion which occurs when other objects partially or fully conceal the target object. This study presents the implementation of the Occlusion Control Block (OCB), which is divided into two sub-blocks: the Occlusion Awareness Block (OAB) and the Estimation Block (EB). The Occlusion Awareness Block (OAB) is integrated with object detection, object size control, target object foreground detection, object velocity control, and the calculation of occlusion duration. EB estimates the target object location sequential frames under occlusion scenario. OCB is integrated to OSTRACK object tracker model as an occlusion handler block, which is checked against other object tracing algorithms through the experiments with LaSOT, LaSOT(Ext), TrackingNet, and GOT10k. The results demonstrate that the integration of our novel OCB design significantly enhances the occlusion handling performance of the base object tracking model. Additionally, the OCB block does not require any dataset training and can be easily integrated into any object tracking algorithm.