00000ctm a22000003i 4500 UP-99796217612614704 Buklod 20230215105418.0 m |o d | ta 171006s xx d r |||| u| (iLib)UPD-00344412941 DENGII rda eng LG 995 2017 E64 D45 Del Carmen, Dale Joshua R. author. Analysis and design of vision-based multiple vehicle tracking systems for traffic monitoring a thesis by Dale Joshhua R. Del Carmen ; Rhandley D. Cajote, thesis adviser. Quezon City College of Engineering, University of the Philippines Diliman 2017. ix, 61 leaves color illustrations 28 cm text rdacontent unmediated rdamedia volume rdacarrier Thesis (M.S. Electrical Engineering)--University of the Philippines Diliman July 2017. F - no patentable invention or creation, not for personal publication and no confidential information. Yes - available to the general public. Multiple-target tracking is one of the main components of traffic monitoring systems directly responsible for measuring traffic information. However, evaluation of these vehicle tracking systems varies from one to the other and are often incomparable due to different metrics and lack of standard data. A comparative framework is created for the evaluation of different online Vehicle tracking Systems on varying traffic scenes. This work focuses on the assessing the viability of online multiple vehicle tracking systems for use in real-time traffic monitoring. The general tracking framework uses background subtraction for real time vehicle detection and various blob-based appearance models for real-time multiple vehicle tracking. The results show that among the blob-based vehicle tracking systems explored, the state-based system performs best in terms of multiple object tracking accuracy (MOTA) of 37.4 percent, multiple object tracking performance (MOTP) of 70.9 percent), and processing speed (44.1 fps). Meanwhile, the feature-based system performs bcst in terms of traffic monitoring, with a relative count and speed accuracy of 90.3 percent, and 83.6 percent, respectively; the main drawback being its processing speed (10.5fps). This shows that commonly used metrics such as MOTA and MOTP are not necessarily reflective of traffic monitoring performance, particularly in terms of vehicle count accuracy. For blob-based tracking systems, in general, track identity Switching is identified to greatly affect the vehicle count accuracy, particularly in terms of count precision. Traffic monitoring. Vehicle detectors. Tracking (Engineering) Cajote, Rhandley D. adviser. Thesis FI UP UPD DENG-II LG 995 2017 E64 D45 Thesis