摘要
Transformer是一种基于自注意力机制的深度神经网络,全局建模能力突出。卷积神经网络(Convolutional Neural Network,CNN)在局部建模方面更有效,二者各具特色,将其组合应用是必然趋势。本文简述了深度学习目标检测算法,介绍了Transformer网络模型。重点以皮革表面图像为研究对象,构建皮革表面瑕疵图像数据集,基于Transformer和CNN构建皮革表面瑕疵自动检测模型。选择了相关硬件设备并介绍了系统软件设计,经过测试证实该模型对于皮革表面瑕疵的检测效果较好,检测准确率在910%以上,有助于皮革瑕疵检测工作的创新改进。
Transformer is a kind of deep neural network based on self-attention mechanism,which has outstanding global modeling ability.Convolutional Neural Network(CNN)is more effective in local modeling,and the two have their own characteristics,and combining them is an inevitable trend.In this paper,deep learning object detection al⁃gorithm and Transformer network model were briefly introduced.Taking leather surface image as the research object,the leather surface defect image dataset was constructed,and an automatic leather surface defect detection model was constructed based on Transformer and CNN.The hardware equipment was selected and the software design of the sys⁃tem was introduced.The test results show that the model has a good detection effect on leather surface defects,and the detection accuracy is more than 91.0%,which is helpful to the innovation and improvement of leather defect de⁃tection.
作者
王玉芳
朱琛
陈江萍
WANG Yufang;ZHU Chen;CHEN Jiangping(Shaanxi Fashion Engineering University,Xi'an 712046,China;Xi'an Zhongtie Railway Transportations Co.,Ltd.,Xi'an 710038,China)
出处
《中国皮革》
CAS
2024年第6期28-31,36,共5页
China Leather
基金
陕西省教育厅项目(23JK0301)
陕西服装工程学院项目(2023XKZ59)。