Scholarlink

<p>Road anomalies such as potholes and uneven surfaces create accident-prone zones that should be properly maintained to develop an easily scalable and intelligent transportation system. This paper presents a mobile crowd sensing system for detecting and spatially mapping such road anomalies. The system leverages a deep learning-based object detection framework, including models such as Mask R-CNN, YOLOv3, and SSD, to analyze images captured by users through a dedicated mobile application. These anomalies are geo-tagged and visualized on an interactive map, providing critical information to road maintenance authorities and commuters. Among the models evaluated, Mask R-CNN with a ResNet-101 backbone achieved the highest performance with a mean Average Precision of 74.6% and a Log-Average Miss Rate of 35.6% at IoU 0.5. The mobile application enables real-time image acquisition, GPS-based tagging, and seamless data transfer to the cloud for processing and storage. The proposed approach offers a scalable, cost-effective, and participatory mechanism to improve road safety and infrastructure planning.</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

Mobile crowd sensing for road anomaly detection using deep learning

  • 期刊论文

  • 发表日期：

    2025年10月27日

查看全文
  • Swadesh Jana,
  • Asif Iqbal Middya,
  • Sarbani Roy

摘要

Road anomalies such as potholes and uneven surfaces create accident-prone zones that should be properly maintained to develop an easily scalable and intelligent transportation system. This paper presents a mobile crowd sensing system for detecting and spatially mapping such road anomalies. The system leverages a deep learning-based object detection framework, including models such as Mask R-CNN, YOLOv3, and SSD, to analyze images captured by users through a dedicated mobile application. These anomalies are geo-tagged and visualized on an interactive map, providing critical information to road maintenance authorities and commuters. Among the models evaluated, Mask R-CNN with a ResNet-101 backbone achieved the highest performance with a mean Average Precision of 74.6% and a Log-Average Miss Rate of 35.6% at IoU 0.5. The mobile application enables real-time image acquisition, GPS-based tagging, and seamless data transfer to the cloud for processing and storage. The proposed approach offers a scalable, cost-effective, and participatory mechanism to improve road safety and infrastructure planning.