A multiple object tracking system applied to insect behavior

Abstract

Segmentation and tracking of multiple objects is an extensively researched eld among Image Sciences. Multiple object tracking is, in general, a very hard problem due to the great number of potencial issues that might arise (such as the sudden movement of the tracked objects, changes on the appearence of either the tracked objects or the background scene, bad-quality frames, occlusion between an object and the scene or between multiple objects, or camera movement). Normally, object tracking is performed on applications that require the object locations to perform calculations later. This paper describes the research, design and development of a system created in order to track multiple insects on recorded videos.