基于云边协同的钢板表面缺陷检测研究与应用

Research and Application of Steel Plate Surface Defect Detection Based on Cloud Edge Collaboration

为了解决钢板表面缺陷检测准确率低、误报率高以及现场应用困难的问题,提出一种基于深度学习的热轧钢板表面缺陷检测方法以及部署方案。选取Faster RCNN目标检测模型对数据预处理并训练、优化模型,以提升识别精度和效率。设计云边协同部署方案,将缺陷检测模型应用部署在现场,以减轻视频传输造成的宽带压力、缩短识别结果延时,满足了企业需求。

In order to solve the problems of low accuracy,high false alarm rate and difficult application of surface defect detection of hot rolled steel plate,the method and deployment scheme of surface defect detection of hot rolled steel plate based on deep learning are proposed.The Fast RCNN target detection model is selected to preprocess the data,train and optimize the model to improve the recognition accuracy and efficiency.Designing the scheme of edge-cloud cooperation deployment and deploying the defect detection model application on the site which reduce the broadband pressure caused by video transmission and shorten the delay of identification result.