摘要
在塔吊事故频发的背景下,将阐述几种当前主流的基于卷积神经网络的目标识别算法应用于塔吊安全监督管理的可行性。将当前主流的目标识别YOLOv3算法、Faster-RCNN算法和SSD算法应用在塔吊的裂缝识别上从而降低塔吊事故的发生,通过比较分析这三种算法在塔吊裂缝识别上的优缺点,并进一步提出下一步的改进方向来更好的针对塔吊安全管理。
Tower crane crack identification is a complex problem.In recent years,tower crane safety accidents frequently occur.Therefore,it is urgent to enhance the ability of tower crane crack detection.In this paper,several current mainstream target recognition algorithms based on convolutional neural networkare described,and the current mainstream target recognition algorithms YOLOv3,Faster-RCNN and SSD are applied to the crack recognition of tower crane.By comparing and analyzing the advantages and disadvantages of these three algorithms in the crack recognition of tower crane,the next improvement direction is put forward.
作者
黄宏安
陈国栋
张神德
HUANG Hongan;CHEN Guodong;ZHANG Shende(Department of Physics and Information Engineering, Fuzhou University ,Fuzhou 350108,China)
出处
《佳木斯大学学报（自然科学版）》
CAS
2021年第1期13-16,共4页
Journal of Jiamusi University：Natural Science Edition
基金
基于车联网云平台的交通违章自动识别关键技术及应用研发(2018H0018)。
关键词
塔吊
裂缝检测
深度学习
卷积神经网络
tower crane
crack detection
deep learning
convolutional neural network
