Scholarlink

The recognition, detection, and classification of traffic signs are extremely beneficial to intelligent transportation systems as they enhance driver safety and compliance with traffic regulations. The study explores traffic sign detection and classification using YOLOv8x and SEResNet-101x deep learning models on the GTSRB and GTSDB datasets. It has a total of 43 classes of traffic signs with a broad variety that will expand the model’s potential to classify specific characteristics. Experimentations on the efficiency of the model are made using image resolutions of 224 x 224 pixels. This analysis compared the performance of YOLOv8x in real-time terms to detect traffic signs and confirmed that the model has demonstrated a satisfactory level of accuracy in recognizing a variety of traffic signs when compared with other models. This work significantly advances the ongoing progress in traffic sign detection and classification tasks by highlighting the utility of YOLOv8x and SEResNet-101x in real-world scenarios and showcasing their potential to enhance driving experience and increase road safety. With a mean accuracy (mAP) of 83.6% for object detection and 99.67% for object classification on GTSRB and GTSDB, the model represents a significant advancement in artificial intelligence for improving traffic safety.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhanced Traffic Sign Detection and Classification Using YOLOv8x and SEResNet-101x Model: A Deep Learning Approach

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Aryan,
  • A. Jackulin Mahariba,
  • Bhavin Joshi,
  • Devansh Singh,
  • Rajveer Singh Chhabra

摘要

The recognition, detection, and classification of traffic signs are extremely beneficial to intelligent transportation systems as they enhance driver safety and compliance with traffic regulations. The study explores traffic sign detection and classification using YOLOv8x and SEResNet-101x deep learning models on the GTSRB and GTSDB datasets. It has a total of 43 classes of traffic signs with a broad variety that will expand the model’s potential to classify specific characteristics. Experimentations on the efficiency of the model are made using image resolutions of 224 x 224 pixels. This analysis compared the performance of YOLOv8x in real-time terms to detect traffic signs and confirmed that the model has demonstrated a satisfactory level of accuracy in recognizing a variety of traffic signs when compared with other models. This work significantly advances the ongoing progress in traffic sign detection and classification tasks by highlighting the utility of YOLOv8x and SEResNet-101x in real-world scenarios and showcasing their potential to enhance driving experience and increase road safety. With a mean accuracy (mAP) of 83.6% for object detection and 99.67% for object classification on GTSRB and GTSDB, the model represents a significant advancement in artificial intelligence for improving traffic safety.