Recently, object recognition in road scenes is attracting interest to the necessity of Advanced Safety Vehicle (ASV). In order to protect Vulnerable Road Users (VRUs) from dangerous vehicles, it is important to identify moving objects quickly and accurately. Motorcycles, one of the dangerous vehicles, are often mistakenly recognized as bicycles because they are similar to bicycles. Motorcycles, however, are faster than bicycles and must be properly distinguished from bicycles considering the danger of collision. According to research on bicycle recognition, Two-Dimensional Histogram of Oriented Gradient (2DHOG) with Variable Window Approach (VWA) shows high accuracy and high processing speed. In this study, 2DHOG with VWA was applied to mixed data including bicycle and motorcycle as Efficient Object Recognition, and the recognition experiment was conducted to confirm whether they can distinguish them. Experimental results distinguished bicycles and motorcycles at a rate of over 90%.


    Zugriff

    Zugriff prüfen

    Verfügbarkeit in meiner Bibliothek prüfen

    Bestellung bei Subito €


    Exportieren, teilen und zitieren



    Titel :

    Efficient Object Recognition with Multi-Directional Features in Urban Scenes


    Beteiligte:


    Erscheinungsdatum :

    01.09.2017


    Format / Umfang :

    1131339 byte




    Medientyp :

    Aufsatz (Konferenz)


    Format :

    Elektronische Ressource


    Sprache :

    Englisch





    Features for Recognition: Viewpoint Invariance for Non-Planar Scenes

    Vedaldi, A. / Soatto, S. / IEEE | British Library Conference Proceedings | 2005



    Efficient Radar Deep Temporal Detection in Urban Traffic Scenes

    Guo, Zuyuan / Wang, Haoran / Yi, Wei et al. | IEEE | 2022