In recent years, traffic safety is the main focus of transport agencies in New Zealand. Due to expansions of cities, the extent of infrastructure-based recordings such as CCTVS grows rapidly. Computer vision can play a critical role in using the data and converting it into useful information. In this paper, we apply the latest deep learning technology to detect and classify traffic users including pedestrians and vehicles. The objects are tracked using a Kalman filter, considering the class with the highest probability in the defined estimation error. We calculate a transformation matrix to map recorded video into a top-down view and measure the speed. Associations between pedestrians and a closest approaching vehicle are considered to predict possible crash points. The elapsed time until pedestrians and vehicles reach a crash point is regarded as the risk time in this context. Obtained results can be used in various traffic safety analysis and decisions.


    Zugriff

    Zugriff prüfen

    Verfügbarkeit in meiner Bibliothek prüfen

    Bestellung bei Subito €


    Exportieren, teilen und zitieren



    Titel :

    Surveillance-based Collision-time Analysis of Road-crossing Pedestrians


    Beteiligte:
    Moayed, Zahra (Autor:in) / Chien, Hsiang-Jen (Autor:in) / Zhang, Derek (Autor:in) / Klette, Reinhard (Autor:in)


    Erscheinungsdatum :

    2019-10-01


    Format / Umfang :

    2087807 byte




    Medientyp :

    Aufsatz (Konferenz)


    Format :

    Elektronische Ressource


    Sprache :

    Englisch



    Analysis of Pedestrians’ Road Crossing Behavior, in Social Groups

    Barón, Leidy / Faria, Susana / Sousa, Emanuel et al. | Transportation Research Record | 2023

    Freier Zugriff


    Automobile longitudinal active collision avoidance control method for pedestrians crossing road based on urban working conditions

    GUO JIAN / LUO MAASIYANG / ZHANG MINGJIA | Europäisches Patentamt | 2022

    Freier Zugriff

    Young Pedestrians' Mobile Phone Use While Crossing the Road

    Juliane Stark / Katja Ruzsicska / Thomas Wiesmann | DOAJ | 2022

    Freier Zugriff