Colloquium "Improving fleet management systems by computer vision"

Abstract:

Fleet management systems monitor hundreds of vehicles by relying on GPS and other simple sensors. Such systems experience challenges in cases of GPS glitches and reduced GPS coverage. Additionally they can provide only limited information about stationary vehicles. This talk will show how these problems can be alleviated by augmenting the GPS information with a short descriptor of an image captured by an on-board camera. The descriptor will allow the server to recognize various scene types by image classification and subsequently improve the quality of service to the clients. Existing state-of-the-art solutions will be reviewed and some ways for adapting them to the problem at hand will be demonstrated. Preliminary experimental evaluation of the proposed methods, and their comparison to the state-of-the-art will be presented using a large dataset specifically designed and collected for this purpose.

CV:
Ivan Sikirić received his M. Eng. (dipl. ing.) degree in Computing in 2007, from the Faculty of Electrical Engineering and Computing, University of Zagreb. He is currently enrolled in the PhD programme of the same Faculty, working towards a PhD in Computing. Since 2008, he is employed in Mireo d.d., where he is working on fleet management systems.  He has co-authored five papers published in the proceedings of international conferences. His research interests include representation and classification in computer vision.