河北工业科技2025，Vol.42Issue(3)：248-257,10.DOI:10.7535/hbgykj.2025yx03006
轻量化U-net模型在钢筋直径测量中的应用研究
Application research on a lightweight U-net model in rebar diameter measurement
摘要
Abstract
In order to solve the problems of low efficiency of traditional manual detection in the acceptance of reinforcement engineering,large measurement errors and even false detection and missing detection caused by human factors in the detection process,a reinforcement diameter measurement method based on an improved lightweight U-net model was proposed.Firstly,a large number of rebar images were collected and a self-made dataset of rebar images was constructed.The MobileNetV3 Block module and CA(coordinated attention)attention mechanism were introduced to improve the classical U-net model.Then,the improved U-net model was trained based on the self-made data set.After training,the test image was imported into the model for segmentation experiment and diameter measurement experiment.The experimental results show that the intersection over union(IoU)of the improved U-net model in the reinforcement image segmentation task is 0.979 5,the model size is only 18.03 MB,and the total average error of diameter measurement experiment is 0.207 mm.The improved model performs well in steel image segmentation,with high detection accuracy and low calculation cost,providing a new technical path for steel image segmentation,and having a certain application prospect in the field of steel image automatic processing and analysis.关键词
土木建筑工程测量/钢筋直径测量/图像分割/改进U-net模型/CA注意力机制Key words
civil engineering survey/rebar diameter measurement/image segmentation/improved U-net model/CA(coordi-nate attention)attention mechanism分类
土木建筑引用本文复制引用
张学辉,于站海,田学昭,安军海,刘新军..轻量化U-net模型在钢筋直径测量中的应用研究[J].河北工业科技,2025,42(3):248-257,10.基金项目
河北省自然科学基金(E2019208150) （E2019208150）
河北科技大学教育教学改革研究与实践项目(2022-YB02) （2022-YB02）
