北京邮电大学学报

  • EI核心期刊

北京邮电大学学报 ›› 2025, Vol. 48 ›› Issue (1): 73-78.

• 论文 • 上一篇    下一篇

基于边缘智能的田间道路缺陷检测方法

陈泽宇1,  古 月1,  程思瑶1,2,3,  冯国惠4   

  1. 1. 哈尔滨工业大学 计算机科学与技术学院;

    2. 哈尔滨工业大学 智慧农场技术与系统全国重点实验室;

    3. 哈尔滨工业大学 中国移动 5G 应用创新联合研究院; 4. 北大荒农垦集团有限公司 北大荒信息有限公司

  • 收稿日期:2023-10-26 修回日期:2024-02-05 出版日期:2025-02-26 发布日期:2025-02-25
  • 通讯作者: 程思瑶 E-mail:csy@hit.edu.cn
  • 基金资助:
    黑龙江省重点研发计划项目; 黑龙江省“揭榜挂帅”科技攻关项目

Field Road Defect Detection Method Based on Edge Intelligence

CHEN Zeyu1,  GU Yue1,  CHENG Siyao1,2,3,  FENG Guohui4   

  1. 1. School of Computer Science and Technology, Harbin Institute of Technology;

    2. National Key Laboratory of Smart Farm Technologies and Systems, Harbin Institute of Technology;

    3. China Mobile 5G Application Innovation Joint Research Institute, Harbin Institute of Technology; 4. Beidahuang Information Company Limited, Beidahuang Group Company Limited

  • Received:2023-10-26 Revised:2024-02-05 Online:2025-02-26 Published:2025-02-25

摘要: 在智慧农机的自主作业任务中,实时的田间道路缺陷检测是确保农机安全工作的关键。但现有道路缺陷检测技术在农业领域研究尚浅,也缺乏针对田间道路的数据集。为得到针对田间道路的缺陷检测模型,首先使用常规道路坑洼数据集训练模型,再利用模拟田间道路坑洼数据集对上述模型进行迁移学习。而为了解决模型在迁移学习后检测精度下降的问题,通过在YOLOv5s网络架构中引入注意力机制来提升网络精度,使田间道路缺陷检测模型检测精度达到83.15%,满足了田间道路缺陷检测的精度要求。为了验证缺陷检测模型的边缘性能,将模型部署到JetsonNano上进行模拟实验。为达到田间道路缺陷检测模型在边缘端的实时检测要求,通过TensorRT技术对模型优化和压缩,使得坑洼检测速度由396ms/帧提升至157ms/帧。

关键词: 缺陷检测 , 迁移学习 ,  深度学习 , 轻量模型设计 , 边缘部署

Abstract:  In the autonomous operation tasks of smart agricultural machinery, real-time field road defect detection is considered essential to ensure the safe operation of the machinery. However, existing road defect detection technologies have been minimally explored in the agricultural domain, and a dedicated dataset for field roads has yet to be developed. To obtain a defect detection model specific to field roads, a model was initially trained on a conventional road pothole dataset, and then followed by transfer learning using a simulated field road pothole dataset. To address the issue of reduced detection accuracy after transfer learning, an attention mechanism was introduced into the YOLOv5s network architecture to enhance the network’s precision, achieving an accuracy of 83.15% for defect detection in field road scenarios and satisfying its accuracy requirements. To verify the edge performance of the defect detection model, it was deployed onto the Jetson Nano for simulation experiments. To meet the real-time detection requirements of the field road defect detection model on edge devices, TensorRT was used to optimize and compress the model, and the pothole detection speed was improved from 396 milliseconds per frame to 157 milliseconds per frame.

Key words:  defect detection, transfer learning, deep learning, lightweight model design, edge deployment

中图分类号: