In order to address the issue of the high frequency and severity of rear-end collisions in traffic accidents and to reduce the occurrence rate of such collisions, this study proposes a real-time vehicle following warning method. The method focuses on the vehicles in the same lane as the subject of research. It involves building a YOLOv7 object recognition model to detect vehicles and traffic signs in the video stream captured by the onboard camera. Additionally, visual distance measurement is utilized to obtain distance information from the subject vehicle to the leading vehicle in the same lane. Furthermore, the driving status information of the leading vehicle is acquired through vehicle-to-vehicle communication. By employing a vehicle following safety distance model, the current safe following distance is calculated, and thus, the real-time vehicle following warning is achieved. The results show that the improved YOLOv7 recognition model performs significantly better than other models in the YOLO series, with a detection time improvement of over 70%, indicating excellent performance. The proposed method enhances the safety of vehicles during the driving process while ensuring driving efficiency, effectively reducing the occurrence of rear-end collisions.


    Zugriff

    Zugriff prüfen

    Verfügbarkeit in meiner Bibliothek prüfen

    Bestellung bei Subito €


    Exportieren, teilen und zitieren



    Titel :

    Research on Vehicle Warning Method Based on Improved YOLOv7


    Beteiligte:
    Ma, Ronggui (Autor:in) / Dong, Shihao (Autor:in) / Zhai, Ziyang (Autor:in)


    Erscheinungsdatum :

    2023-08-18


    Format / Umfang :

    2043928 byte




    Medientyp :

    Aufsatz (Konferenz)


    Format :

    Elektronische Ressource


    Sprache :

    Englisch



    Freeway vehicle detection on improved YOLOv7

    Gong, Shaojie / Mao, Kun / Yuan, Haifeng et al. | SPIE | 2023




    Improved YOLOv7 Target Detection Algorithm Based on UAV Aerial Photography

    Zhen Bai / Xinbiao Pei / Zheng Qiao et al. | DOAJ | 2024

    Freier Zugriff