Scholarlink

<p>Accurate and reliable road surface quality data enable travelers to select optimal routes, which is essential for efficient transportation. In this context, monitoring road surface conditions (RSC) using smartphone motion sensor data has emerged as a key area of research in the transportation sector. This paper introduces a data augmentation framework designed to generate synthetic time-series training data for 3D hybrid deep learning models, enabling accurate and reliable road anomaly detection. Through evaluating performance using the averaging ensemble of the 3D models, the efficacy of the proposed data augmentation techniques is verified. The experimental results have demonstrated the significant potential of data augmentation in enhancing the accuracy of road anomaly detection.</p>

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

Enhancing Road Anomaly Detection with Data Augmentation and 3D Hybrid Deep Learning Models

  • 期刊论文

  • 发表日期：

    2026年6月15日

查看全文
  • Yacine Kabir,
  • Abdelkader Hadj-Attou,
  • Farid Ykhlef

摘要

Accurate and reliable road surface quality data enable travelers to select optimal routes, which is essential for efficient transportation. In this context, monitoring road surface conditions (RSC) using smartphone motion sensor data has emerged as a key area of research in the transportation sector. This paper introduces a data augmentation framework designed to generate synthetic time-series training data for 3D hybrid deep learning models, enabling accurate and reliable road anomaly detection. Through evaluating performance using the averaging ensemble of the 3D models, the efficacy of the proposed data augmentation techniques is verified. The experimental results have demonstrated the significant potential of data augmentation in enhancing the accuracy of road anomaly detection.