{"id":18732,"date":"2014-06-18T00:00:00","date_gmt":"2014-06-18T00:00:00","guid":{"rendered":"https:\/\/carsp.ca\/?p=18732"},"modified":"2022-10-30T00:54:28","modified_gmt":"2022-10-30T00:54:28","slug":"automated-classification-based-on-video-data-at-intersections-with-heavy-pedestrian-and-bicycle-traffic-methodology-and-application","status":"publish","type":"post","link":"https:\/\/carsp.ca\/en\/presentations-and-papers\/2014-cmrsc-ccmsr-xxiv-vancouver\/automated-classification-based-on-video-data-at-intersections-with-heavy-pedestrian-and-bicycle-traffic-methodology-and-application\/","title":{"rendered":"Automated Classification Based on Video Data at Intersections with Heavy Pedestrian and Bicycle Traffic: Methodology and Application"},"content":{"rendered":"Author(s): Sohail Zangenehpour, Luis F. Miranda-Moreno, Nicolas Saunier
\r\nStudent Paper Competition: 2nd Place<\/span>\n

Slidedeck Presentation:<\/h2>\n

6C Zangenehpour Automated Classification in Traffic Video at Intersections<\/a><\/p>\n

<\/div>\n

Abstract:<\/h2>\n

Pedestrians and cyclists are amongst the most vulnerable road users. Pedestrian and cyclist
\ncollisions involving motor-vehicles result in high injury and fatality rates for these two modes.
\nData for pedestrian and cyclist activity at intersections such as volumes, speeds, and space-time
\ntrajectories are essential in the field of transportation in general, and road safety in particular.
\nHowever, automated data collection for these two road user types remains a challenge. Due to
\nthe constant change of orientation and appearance of pedestrians and cyclists, detecting and
\ntracking them using video sensors is a difficult task. This paper presents a method based on
\nHistogram of Oriented Gradients to extract features of an image box containing the tracked
\nobject and Support Vector Machine to classify moving objects in crowded traffic scenes. Moving
\nobjects are classified into three categories: pedestrians, cyclists, and motor vehicles. The
\nproposed methodology is composed of three steps: i) detecting and tracking each moving object
\nin video data, ii) classifying each object according to its appearance in each frame, and iii)
\ncomputing the probability of belonging to each class based on both object appearance and its
\nspeed. For the last step, Bayes\u2019 rule is used to fuse appearance and speed in order to predict
\nthe object category. Using various video datasets collected in different intersections, the
\nmethodology was developed and tested. The developed methodology shows an overall
\nclassification accuracy of more than 90 %. However, the classification accuracy varies across
\nmodes and is highest for vehicles and lower for pedestrians and cyclists. The applicability of the
\nproposed methodology is illustrated using a simple case study to analyse cyclist-vehicle conflicts
\nat intersections with and without cycle tracks.<\/p>\n

<\/div>Sohail Zangenehpour, Luis F. Miranda-Moreno, Nicolas Saunier
\r\nStudent Paper Competition: 2nd Place<\/span>\n

R\u00e9sum\u00e9 :<\/h2>\n

Les pi\u00e9tons et les cyclistes sont parmi les usagers de la route les plus vuln\u00e9rables. Les collisions
\ndes pi\u00e9tons et des cyclistes avec des v\u00e9hicules motoris\u00e9s entra\u00eenent des taux \u00e9lev\u00e9es de
\nmortalit\u00e9 et de blessures graves pour ces deux modes. Des donn\u00e9es sur l\u2019activit\u00e9 des pi\u00e9tons et
\ndes cyclistes aux carrefours, telles que des d\u00e9bits, des vitesses et des trajectoires, sont
\nessentielles dans le domaine des transports et pour la s\u00e9curit\u00e9 routi\u00e8re en particulier.
\nCependant, collecter automatiquement des donn\u00e9es pour ces deux types d\u2019usagers de la route
\nrepr\u00e9sente encore un d\u00e9fi. D\u00e9tecter et suivre les pi\u00e9tons et les cyclistes \u00e0 l\u2019aide de capteurs
\nvid\u00e9o est une t\u00e2che difficile \u00e0 cause de leurs changements d\u2019orientation et d\u2019apparence. Cet
\narticle pr\u00e9sente une m\u00e9thode reposant sur le calcul de descripteurs (histogramme de gradients
\norient\u00e9s) pour des r\u00e9gions d\u2019image contenant l\u2019objet suivi et des machines \u00e0 support de vecteurs
\npour classifier les objets en mouvement dans des sc\u00e8nes avec des d\u00e9bits \u00e9lev\u00e9s. Les objets en
\nmouvements sont classifi\u00e9s en trois cat\u00e9gories : pi\u00e9tons, cyclistes et v\u00e9hicules motoris\u00e9s. La
\nm\u00e9thode pr\u00e9sent\u00e9e suit les trois \u00e9tapes suivantes: i) d\u00e9tecter et suivre chaque objet en
\nmouvement dans les donn\u00e9es vid\u00e9o, ii) classifier chaque objet selon son apparence dans
\nchaque image, iii) calculer la probabilit\u00e9 d\u2019appartenir \u00e0 chaque classe d\u2019usager de la route selon
\nl\u2019apparence et la vitesse. La derni\u00e8re \u00e9tape repose sur la r\u00e8gle de Bayes pour fusionner les
\ninformations d\u2019apparence et de vitesse dans le but de pr\u00e9dire la cat\u00e9gorie de l\u2019objet. La m\u00e9thode
\na \u00e9t\u00e9 d\u00e9velopp\u00e9e et test\u00e9e sur plusieurs ensembles de donn\u00e9es vid\u00e9o. Le taux global de bonne
\nclassification est sup\u00e9rieur \u00e0 90 %. La performance varie cependant selon les modes, du plus
\n\u00e9lev\u00e9 pour les v\u00e9hicules motoris\u00e9s au plus bas pour les pi\u00e9tons et les cyclistes. La m\u00e9thode est
\nappliqu\u00e9e pour illustration sur un cas d\u2019\u00e9tude simple d\u2019analyse des conflits v\u00e9hicules-pi\u00e9tons \u00e0
\ndes carrefours avec et sans piste cyclable.","protected":false},"excerpt":{"rendered":"

Sohail Zangenehpour, Luis F. Miranda-Moreno, Nicolas Saunier
\nStudent Paper Competition: 2nd Place<\/span><\/p>\n","protected":false},"author":163,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"_kad_post_transparent":"default","_kad_post_title":"default","_kad_post_layout":"default","_kad_post_sidebar_id":"","_kad_post_content_style":"default","_kad_post_vertical_padding":"default","_kad_post_feature":"","_kad_post_feature_position":"","_kad_post_header":false,"_kad_post_footer":false,"_kad_post_classname":"","footnotes":""},"categories":[128,346],"tags":[],"class_list":["post-18732","post","type-post","status-publish","format-standard","hentry","category-2014-cmrsc-ccmsr-xxiv-vancouver","category-research-and-evaluation"],"acf":[],"_links":{"self":[{"href":"https:\/\/carsp.ca\/en\/wp-json\/wp\/v2\/posts\/18732","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/carsp.ca\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/carsp.ca\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/carsp.ca\/en\/wp-json\/wp\/v2\/users\/163"}],"replies":[{"embeddable":true,"href":"https:\/\/carsp.ca\/en\/wp-json\/wp\/v2\/comments?post=18732"}],"version-history":[{"count":3,"href":"https:\/\/carsp.ca\/en\/wp-json\/wp\/v2\/posts\/18732\/revisions"}],"predecessor-version":[{"id":19787,"href":"https:\/\/carsp.ca\/en\/wp-json\/wp\/v2\/posts\/18732\/revisions\/19787"}],"wp:attachment":[{"href":"https:\/\/carsp.ca\/en\/wp-json\/wp\/v2\/media?parent=18732"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/carsp.ca\/en\/wp-json\/wp\/v2\/categories?post=18732"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/carsp.ca\/en\/wp-json\/wp\/v2\/tags?post=18732"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}