Multi-objective Tracking Applied to Bat Populations

Multiple target tracking is one of the great challenges faced by the computer vision community in last years. This paper presents a solution developed in order to track bats in a clutter environment to account the population of their colony. The algorithm is able to start detections, treat wrong or lost detections and process the detections in progress. Tracked targets have their next state estimated by Kalman Filter usage. As new measurements are performed, the algorithm can identify which of them are considered noise, which are new targets and which are new states of a previously detected target. A 3D viewer was also implemented in order to help the analysis of the tracked flights by researchers in areas like biodiversity and biology. The aim of this paper is to present the operation of the developed system, collaborate with other researchers working with tracking of multiple objects and make society aware of the importance of preserving the environment, exposing some of the consequences of changing its natural characteristics. The proposed algorithm showed amazing results in the test stages, reaching to overcome the current state of the art.