摘要
为解决传统裂缝检测工作效率低、危险系数高、耗时较长及检测精度低等问题,提出一种大型建筑物远距离裂缝检测智能平台方案,该方案基于深度学习和图像处理建立裂缝检测算法模型,采集图像后上传至云端,以用户终端设备为载体,通过微信小程序和后端服务器的交互获取裂缝相关信息,服务器实现裂缝检测模型的实时检测,裂缝数据存储于云端;经过对算法部分的图像分类模型与图像分割模型进行测试,分类模型精确率、召回率、F值均超过92%,分割模型平均交并比MIoU超过89%,满足检测需要。经验证该方案切实可行。
In order to solve the traditional crack detecting low working efficiency,high risk,takes longer and low accuracy of detection problem,put forward a kind of large buildings long crack detection intelligence platform solutions,the scheme is based on deep study establishes a model of crack detection algorithm and image processing,after acquisition images uploaded to the cloud,as user terminal equipment as the carrier,The fracture related information is obtained through the interaction between the Wechat applet and the back-end server.The server realizes real-time detection of the fracture detection model,and the fracture data is stored in the cloud.After testing the image classification model and image segmentation model in the algorithm part,the accuracy rate,recall rate and F value of the classification model are more than 92%,and the average intersection union ratio of the segmentation model is more than 89%,which meets the needs of detection.It is proved that the scheme is feasible.
作者
余欣
许益奖
李薇
石威
刘晓宇
YU Xin;XU Yijiang;LI Wei;SHI Wei;LIU Xiaoyu(School of Mechanical Engineering,Sichuan University,Chengdu 610000,China)
出处
《计算机测量与控制》
2022年第12期70-77,共8页
Computer Measurement &Control
基金
2022年四川省“大学生创新创业训练计划项目”(S202210610375)。
