Scholarlink

With the advancement of autonomous driving systems, traffic sign detection has become increasingly important in the context of intelligent transportation systems (ITS), aiming to reduce transportation challenges and lower traffic accident rates. This study evaluates the performance of the YOLOv5 and YOLOv8 object detection algorithms for traffic sign detection, a critical task in ITS. Both models were tested and compared using a dataset of traffic sign images to assess their precision, recall, mean average precision (mAP), F1 score, and processing speed (FPS). YOLOv8 demonstrated significant improvements over YOLOv5, achieving a precision of 0.928, recall of 0.853, mAP of 0.877, and an F1 score of 0.888, while processing at an average rate of 75 FPS. In comparison, YOLOv5 achieved a precision of 0.839, recall of 0.746, mAP of 0.782, and an F1 score of 0.790 at 49 FPS. The results indicate that YOLOv8 provides not only higher detection accuracy but also significantly faster processing, making it more suitable for real-time traffic sign detection applications. The findings highlight the advancements in YOLOv8’s architecture, particularly its multi-scale feature fusion and anchor-free detection mechanisms, which contribute to its superior performance.

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

Deep Learning-Based Detection and Classification of Traffic Signs for Autonomous Driving Assistance

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Behiye Sahin,
  • Murat Bakirci

摘要

With the advancement of autonomous driving systems, traffic sign detection has become increasingly important in the context of intelligent transportation systems (ITS), aiming to reduce transportation challenges and lower traffic accident rates. This study evaluates the performance of the YOLOv5 and YOLOv8 object detection algorithms for traffic sign detection, a critical task in ITS. Both models were tested and compared using a dataset of traffic sign images to assess their precision, recall, mean average precision (mAP), F1 score, and processing speed (FPS). YOLOv8 demonstrated significant improvements over YOLOv5, achieving a precision of 0.928, recall of 0.853, mAP of 0.877, and an F1 score of 0.888, while processing at an average rate of 75 FPS. In comparison, YOLOv5 achieved a precision of 0.839, recall of 0.746, mAP of 0.782, and an F1 score of 0.790 at 49 FPS. The results indicate that YOLOv8 provides not only higher detection accuracy but also significantly faster processing, making it more suitable for real-time traffic sign detection applications. The findings highlight the advancements in YOLOv8’s architecture, particularly its multi-scale feature fusion and anchor-free detection mechanisms, which contribute to its superior performance.