According to the YOLOv3Tiny Vehile high-speed vehicle real-time detection network and method, the detection efficiency is improved under the condition that the detection precision is guaranteed, and the network and method are particularly suitable for vehicle real-time detection under the conditions that the vehicle speed is high and the vehicle flow is large. And the YOLOv3Tiny Vehile network comprises a trunk feature extraction module and a detection module. The trunk feature extraction module comprises a trunk feature extraction first layer, five dense connection blocks and a trunk feature extraction tail layer. And the trunk feature extraction first layer is connected with the first dense connection block. The five dense connecting blocks, namely the first dense connecting block, the second dense connecting block, the third dense connecting block, the fourth dense connecting block and the fifth dense connecting block, are of the same structure and are connected in sequence. Each dense connection block is composed of three convolutional layers, and the three convolutional layers are connected in a dense connection mode; and the fifth dense connection block is connected with the trunk feature extraction tail layer. The detection module comprises three convolution layers which are connected in sequence.

    本发明提出一种YOLOv3Tiny Vehicle高速车辆实时检测网络及方法,在保证检测精度的情况下提高了检测效率,尤其适用于车辆速度较快、车流量较多情况的车辆实时检测。YOLOv3Tiny Vehicle网络包括主干特征提取模块和检测模块。主干特征提取模块包括主干特征提取首层、五个密集连接块和主干特征提取尾层。主干特征提取首层与第一密集连接块相连。五个密集连接块,即第一密集连接块、第二密集连接块、第三密集连接块、第四密集连接块和第五密集连接块,结构相同、顺序相连;每个密集连接块由三个卷积层构成,这三个卷积层的连接采用密集连接方式;第五密集连接块与主干特征提取尾层相连。检测模块包括顺序相连的三个卷积层。


    Access

    Download


    Export, share and cite



    Title :

    YOLOv3 Tiny Vehile high-speed vehicle real-time detection network and method


    Additional title:

    YOLOv3 Tiny Vehicle高速车辆实时检测网络及方法


    Contributors:
    CUI YANI (author) / REN JIA (author) / HAO QIUSHI (author) / ZHANG YU (author) / YI JIAFU (author) / DING JIE (author) / CHEN MIN (author) / MA XIUXIN (author)

    Publication date :

    2023-06-13


    Type of media :

    Patent


    Type of material :

    Electronic Resource


    Language :

    Chinese


    Classification :

    IPC:    G06V / G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen / G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS



    YOLOv3 tiny vehicle: a new model for real-time vehicle detection

    Ma, Xiuxin / Zhuang, Huaiyu / Deng, Jiaxian et al. | SPIE | 2022

    Free access

    REAL-TIME CURRENT CAR DEBUGGING AND SIMULATION SYSTEM FOR RAILWAY VEHILE

    OHN JUNG GHUN | European Patent Office | 2022

    Free access


    SHIP DETECTION IN MARINE RADAR IMAGES BASED ON A MODIFIED YOLOV3-TINY

    Lei, Y. / Si, L. / Sun, L. et al. | TIBKAT | 2021


    Real-Time Pedestrian Detection and Tracking Based on YOLOv3

    Li, Xingyu / Hu, Jianming / Liu, Hantao et al. | TIBKAT | 2022