In the realm of intelligent vehicles, the evolution of object detection algorithms is of paramount importance. Current deep learning-based methodologies excel in identifying medium to large-sized objects but often falter with smaller entities. A notable research gap exists in integrating key vehicular state data, such as velocity and steering angle, into generally designed object detection frameworks. To bridge these gaps, we present Prior-YOLO, a novel modification of YOLO v8, marked by advanced network structure and refined inference processes. This adaptation includes a dedicated head for small object detection and a new neck component. We have also curated a custom dataset, particularly for algorithm development and validation. This dataset, enriched with critical driving status data, is rigorously annotated to maximize precision. Furthermore, we introduce the Visual Center of Mass Method (VCM), a groundbreaking technique that integrates the driver's focal point to identify the region of interest (ROI) in each image, enhancing auxiliary inference. Our comprehensive experimental evaluations demonstrate significant improvements, with a notable increase of up to 6.04% in Average Precision (AP) and up to 5.44% in Average Precision for Small Objects (APs) across all objects. Most impressively, we observe a 9.24% increase in AP and a 12.59% rise in APs for critical objects, solidifying the effectiveness of Prior-YOLO in intelligent vehicle applications.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    Prior-YOLO: Enhancing Intelligent Vehicle Small Object Detection with Driving Status-Informed YOLOv8


    Contributors:
    Hu, Shuang (author) / Zhao, Baixuan (author) / Ding, Taojun (author) / Jiang, Hao (author) / Hu, Chuan (author) / Zhang, Xi (author)


    Publication date :

    2023-12-15


    Size :

    1765101 byte




    Type of media :

    Conference paper


    Type of material :

    Electronic Resource


    Language :

    English





    SRS-YOLO: Improved YOLOv8-Based Smart Road Stud Detection

    Mao, Guoqiang / Wang, Keyin / Du, Haoyuan et al. | IEEE | 2025


    Enhancing real-time object detection for UAVs using optimized YOLOv8 architecture

    Li, Wenrui / Liu, Na / Zhang, Pizhen et al. | SPIE | 2025


    Intelligent driving interaction and control system based on YOLOv8

    HOU ALIN / ZHANG JUNPENG / XU MINJUN et al. | European Patent Office | 2023

    Free access