Application of ID Assignment Features in Rosebrocks' Simple Object Tracker

Recent advancements in object detection methods give us the means to recognize what kind of object is present in the sensor field of view. Though indeed useful, this may not be enough for some specific application in which we need to distinguish an object from another object, that might belong to the same or different class object. This paper will discuss a good option for a high-speed/low computation cost method for ID assignment that can be used even for objects that belong to the same class. Utilizing the Centroid Tracker concept, Rosebrock applies an ID assignment based on distances between objects in the current and previous frame. This method has shown promising success in complement with object detection to help a computer vision system for tracking specific objects of interest.