| 摘 要: 针对目前道路裂缝检测精度低,难以识别细小裂缝的问题,提出了一种改进的YOLOv9c算法模型,显著提升了道路裂缝的检测效果。首先,在YOLOv9c模型的头部网络中引入了CBAM模块,这不仅提高了模型的特征提取能力,还增加了检测精度;然后,在骨干网络和头部网络中用SCDown模块替换原有的 ADown模块,在降低模型复杂度的同时,提升了检测效率。经过实验,改进的YOLOv9c模型与原模型相比,分别在精度、召回率与mAP上各自提高了2.3%、3.1%与2.7%,模型参数量降低了10%,实现了对道路细小裂缝的高精度和高效率检测。 |
| 关键词: 道路裂缝检测 YOLOv9c 深度学习 CBAM 卷积神经网络 |
|
中图分类号: TP391.4
文献标识码: A
|
|
| Research on Road Crack Recognition Model Based on Deep Learning |
|
XU Zihao, ZHU Xiaohong, ZOU Lanlin
|
(The School of Automobile and Traffic Engineering, Wuhan University of Science and Technology, Wuhan 430070, China)
2011301011@qq.com; zhuxiaohong@wust.edu.cn; zll999-9@163.com
|
| Abstract: To address the current issues of low accuracy in road crack detection and difficulty in identifying fine cracks, this study proposes an improved YOLOv9c algorithm model, which significantly enhances detection performance. Firstly, the CBAM module is introduced into the head network of the YOLOv9c model, improving both feature extraction capability and detection accuracy. Secondly, the original ADown module is replaced with the SCDown module in both the backbone and head networks, reducing model complexity while increasing detection efficiency. Experimental results demonstrate that compared with the original model, the improved YOLOv9c model achieves increases of 2.3% , 3.1% , and 2.7% in precision, recall, and mAP respectively, while reducing model parameters by 10% . This enables high-precision and efficient detection of fine road cracks. |
| Keywords: road crack detection YOLOv9c deep learning CBAM convolutional neural network |
