基于深度学习的混凝土坍落度检测方法

Concrete slump estimation method based on deep learning

随着城市化建设的大幅加快,混凝土作为最重要的建筑材料之一,在生产与施工过程中进行质量控制尤为重要。然而,传统的混凝土工作性能检测方法十分费时费力,并且检测结果的准确性受到人为操作因素的影响较大,不利于工程建设的安全与高效实施。本文提出了一种简便、高效的混凝土坍落度检测方法,采用深度学习对混凝土拌合物图像进行识别,达到快速检测混凝土坍落度的目的。本次采用了ResNet,ResNeXt,DenseNet和MobilenetV3架构进行图像分析,经过数据构建、模型训练和应用测试,分析结果证明了计算机视觉方法在混凝土坍落度检测过程中的有效性和准确性,促进建筑工程行业的数字化智能化技术应用进一步发展。

In the context of accelerated urbanization,the quality of concrete,as one of the most extensively used building materials,is crucial for the safety and durability of construction projects.However,traditional methods for inspecting concrete workability are timeconsuming and often limited in accuracy due to improper operations.This research explores the potential application of deep learning technologies to enhance the efficiency and accuracy of concrete workability inspections.Specifically,the study introduces the use of ad⁃vanced neural network architectures,such as ResNet,ResNeXt,DenseNet,and MobilenetV3 models,implemented to detect and evalu⁃ate concrete slump values.Experimental results demonstrate that these models significantly improve the speed,accuracy,and cost-ef⁃fectiveness of concrete slump inspections.This study not only contributes to the technical fields of automated inspection and deep learn⁃ing but also supports sustainable development in the construction industry by advancing its technological framework.