| 摘要: |
| 由于道路缺陷呈现多尺度的特征，导致检测的准确度不高。为了改进这种状况，设计了一种YOLOv8-MCMA道路缺陷检测模型。首先，使用MobileViT结构，能够保证降低参数的情形下还保持较高的识别准确率的优势；其次，采用内容感知的特征重组 (Content-Aware ReAssembly of Features, CARAFE)为上采样模块，致力于检测出更多细小的裂缝图像；设计了多尺度倒置残差注意力模块（Multi-scale Inverted Residual Attention, MIRA），模型对多尺度的特征更加敏感；最后，将颈部的普通卷积换为可变核卷积（Alterable Kernel Convolution ，AKConv），不规则的裂缝信息能够被更加重视，以此来降低检测的失误。实验结果表明，在Road damage detection Dataset、RDD2022_China和Crack-forest dataset数据集上，与YOLOv8n模型相比mAP@0.5分别提高了3.7%、1.4%和2.6%，参数量减少了23%，与其他算法相比有明显的优势，对多尺度的道路缺陷有很强的适应能力。 |
| 关键词: 计算机视觉 目标检测 道路缺陷检测 MobileViT MIRA YOLOv8-MCMA |
| DOI： |
| 投稿时间：2024-09-24修订日期：2024-11-18 |
| 基金项目:辽宁省教育厅科研经费项目(LJKMZ20220838) |
|
| Application Research of Road Defect Detection Based on YOLOv8-MCMA Model |
|
XU Kesheng, SUN Rong
|
| (Dalian Jiaotong University) |
| Abstract: |
| The detection precision for road defects, which come in various sizes has often been insufficient. To tackle this issue, the YOLOv8-MCMA road defect detection model has been specifically designed. Firstly, the model integrates the MobileViT structure, maintaining high recognition accuracy even with a reduced parameter count. Second, it employs the Content-Aware ReAssembly of Features (CARAFE) as an up-sampling module, focusing on the detection of small-scale crack images. Additionally, a Multi-scale Inverted Residual Attention module (MIRA) is introduced to enhance the model's sensitivity to features across different scales. Finally, the traditional convolution in the model's neck is replaced with an Alterable Kernel Convolution (AKConv), which better captures irregular crack information, thus reducing detection errors. The experimental results show that in the Road damage detection Dataset, RDD2022_China and Crack-forest dataset, the value of mAP@0.5 increased by 3.7%, 1.4% and 2.6%, respectively, compared with the YOLOv8n model. Compared to the YOLOv8 model, there is a 23% reduction in parameter count. The YOLOv8-MCMA model demonstrates significant advantages over other algorithms and shows strong adaptability to detecting multi-scale road defects. |
| Key words: computer vision object detection road defect detection MobileViT MIRA YOLOv8-MCMA |
