There is a rise in car accidents due to human errors on the road. A critical task of self-driving cars that can reduce accidents on the road is traffic sign detection and recognition (TSDR), which is vital in alerting drivers to the presence of traffic signs in advance. This research will separate the proposed deep ensemble learning algorithm into two methods. First, after the traffic scene process, the algorithm will detect the traffic sign as two categories with the YOLOv5s network. Then, process the traffic sign to recognize the traffic sign into seven classes with the MobileNet network. The detection model was trained with the Taiwan Traffic Sign Detection (TTSD) dataset collected from Taiwan roads. The recognition model was trained with the Taiwan Traffic Sign Recognition (TTSR) dataset. The result of the proposed algorithm showed high performance when experimenting with 95.83% accuracy, 87.34% true prediction, and 191.3 milliseconds (ms) of inference time.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    Research on Traffic Sign Detection and Recognition System Using Deep Ensemble Learning




    Publication date :

    2024-06-14


    Size :

    1762589 byte





    Type of media :

    Conference paper


    Type of material :

    Electronic Resource


    Language :

    English



    Traffic Sign Detection and Recognition using Deep Learning

    Oza, Rudri Mahesh / Geisen, Angelina / Wang, Taehyung | IEEE | 2021


    Traffic Sign Detection and Recognition Using Deep Learning

    Kumar, P. Puneeth / Kishen, R. C. / Ravikumar, M. | Springer Verlag | 2022



    Traffic Sign Detection and Recognition Using Ensemble Object Detection Models

    Quasar, Syeda Reeha / Sharma, Rishika / Mittal, Aayushi et al. | Springer Verlag | 2023


    Indian traffic sign detection and recognition using deep learning

    Rajesh Kannan Megalingam / Kondareddy Thanigundala / Sreevatsava Reddy Musani et al. | DOAJ | 2023

    Free access