<  Back to the Polytechnique Montréal portal

Systematic Parameter Optimization and Application of Automated Tracking in Pedestrian-Dominant Situations

Dariush Ettehadieh

Masters thesis (2014)

[img]
Preview
Download (22MB)
Cite this document: Ettehadieh, D. (2014). Systematic Parameter Optimization and Application of Automated Tracking in Pedestrian-Dominant Situations (Masters thesis, École Polytechnique de Montréal). Retrieved from https://publications.polymtl.ca/1661/
Show abstract Hide abstract

Abstract

RÉSUMÉ Les mouvements des piétons et leur modélisation constituent un domaine de recherche de plus en plus actif. Bien qu’encore souvent appliqué à la sécurité par l’élaboration de plans d’évacuation en cas d’urgence, comprendre le mouvement des piétons est un enjeu économique de plus en plus important, notamment pour améliorer l’efficacité des aménagements de transport et des grands centres commerciaux. Cependant, les données existantes — particulièrement au niveau individuel, ou microscopique —sont majoritairement collectées dans des situations expérimentales contrôlées. Elles ne sont donc pas nécessairement représentatives du comportement des piétons dans des situations réelles, particulièrement en tenant compte de la susceptibilité de leur comportement aux facteurs démographiques, psychologiques et nvironnementaux. Cette lacune est due principalement à l’absence de méthodes prouvées pour la détection et le suivi de piétons dans des cas réels, absence qui résulte de la complexité des mouvements piétons et qui persiste malgré l’avancement continu des méthodes automatique d’analyse.----------ABSTRACT Though a wealth of data exists for the characterization of pedestrian movement, a majority of it originates from experimental settings owing to the current state of trackers for real-world scenarios. While these trackers are steadily improving, they remain insufficiently reliable for the accurate, microscopic tracking of individuals, particularly in cases of occlusion or higher density, complex scenes. In this work, the use of evolution algorithms is proposed for the systematic calibration of the parameters of existing trackers in order to further optimize their performance – evaluated by tracking accuracy and precision metrics – in complex cases, with an initial focus on two tracking methods designed for multimodal analysis. This calibration is further aided by the inclusion of additional parameters regulating homography, or specifically the plane to which tracker detections are projected. Three real test cases were used: a) a confined corridor in a public building, b) a subway station entrance during morning rush hour and c) a crosswalk in downtown New York. Results demonstrate a halving of tracking errors over both default and manually-calibrated parameters, as well as a strong correlation in performance between similar cases. These results were consistent over multiple trials and regardless of the starting parameters, strongly implying that the obtained solutions are indeed the global maxima for each scene. For application and validation of the resultant tracks, flow characterization and directional counting are demonstrated, utilizing tools included in the optimization framework.

Open Access document in PolyPublie
Department: Département des génies civil, géologique et des mines
Dissertation/thesis director: Nicolas Saunier and Bilal Farooq
Date Deposited: 01 Apr 2015 14:26
Last Modified: 27 Jun 2019 16:48
PolyPublie URL: https://publications.polymtl.ca/1661/

Statistics

Total downloads

Downloads per month in the last year

Origin of downloads

Repository Staff Only