基于改进Deeplab V3+网络的裂缝检测

Crack Detection Based on Improved Deeplab V3+Network

裂缝是常见的路面损坏现象。由于裂缝容易扩展,对早期裂缝进行检测具有重要的现实意义。传统的人工裂缝检测极其耗时耗力,研究人员已将注意力转向自动裂缝检测。尽管自动裂缝检测在过去几十年中得到了广泛的研究,但由于裂缝的不均匀性和路面环境的复杂性,它仍然是一项具有挑战性的任务。为了解决这些问题,提出一种基于改进的Deeplab V3+网络的路面裂缝分割模型。在原始Deeplab V3+网络的基础上将主干网络替换成resnet-50网络,并对空洞空间卷积池化金字塔模块进行了级联操作。为了验证所提出方法的有效性,在CRACK500数据集上进行了训练和测试。本文方法在CRACK500数据集上的平均交并比(Mean Intersection over Union,MIoU)和平均像素精度(Mean Pixel Accuracy,MPA)达到了0.8315和0.8614,优于原始Deeplab V3+网络的检测结果。

Cracks are the most common pavement damage.Due to the propagation of cracks,the detection of early cracks has important practical significance.Traditional manual crack detection is extremely time-consuming and labor-intensive,and researchers have turned their attention to automatic crack detection.Although automatic crack detection has been extensively studied in the past decades,it remains a challenging task due to the heterogeneity of cracks and the complexity of the pavement environment.To solve these problems,this paper proposes an effective pavement crack segmentation model based on the improved Deeplab V3+network.In this paper,based on the original Deeplab V3+network,the backbone network is replaced by the resnet-50 network,and the convolutional pooling pyramid module of the empty space is cascaded.To verify the effectiveness of the proposed method,we train and test on the CRACK500 dataset.The Mean Intersection over Union(MIoU)and Mean Pixel Accuracy(MPA)of our method on the CRACK500 dataset reach 0.8315 and 0.8614,which are better than the detection results of the original Deeplab V3+network.