YOLO Based Bridge Surface Defect Detection Using Decoupled Prediction
Authors:
- 1 July 2022
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 117-122
Abstract
Bridge surface defect detection is an essential method for evaluating the bridge quality and subsequent repair. Convolutional neural network based intelligent object detectors have been powerful for defect detection in recent years. In this paper, we propose a model based on the YOLO detector for detection. We use the convolutional block attention module, decoupled prediction head, and focal loss function to improve performance. To verify our proposed model, we perform experiments on an open bridge surface defect dataset, and our model can obtain 90.3% mAP50 and 72.8% mAP75. The detection precision of our network is higher than the original YOLOv5.Keywords
Funding Information
- Research and Development
- National Natural Science Foundation of China
This publication has 28 references indexed in Scilit:
- UnitBox2016
- SSD: Single Shot MultiBox Detector2016
- Road crack detection using deep convolutional neural network2016
- You Only Look Once: Unified, Real-Time Object Detection2016
- A review on computer vision based defect detection and condition assessment of concrete and asphalt civil infrastructureAdvanced Engineering Informatics, 2015
- Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation2014
- Learning Feature Representations with K-Means2012
- The Pascal Visual Object Classes (VOC) ChallengeInternational Journal of Computer Vision, 2009
- Image Processing Algorithm for Apple Defect DetectionTransactions of the ASAE, 1989
- A Stochastic Approximation MethodThe Annals of Mathematical Statistics, 1951