摘要
针对皮革缺陷分类存在误判、成本较高及目前关于皮革缺陷的研究主要是针对皮革做缺陷检测,未进行缺陷分类的问题,采用一种参数优化的残差网络来实现皮革缺陷的自动分类。首先通过多层卷积、池化操作进行特征提取,并引入残差模块解决深层网络的梯度消失问题;然后依据所提取特征进行缺陷分类;最后根据皮革数据集优化关键网络参数,使用数据增强方法对数据集进行扩充,有效避免了网络模型因样本不足易产生过拟合的问题。实验结果表明该方法可对皮革缺陷进行有效分类,分类精度达到92.34%。
The current research on leather defects is mainly aimed at defect detection without classification of defect types.Also,manual classification has disadvantages including misjudgment and high cost.To address these issues,a parameter-optimized residual network was presented to realize automatic classification of the leather defects.Firstly,features were extracted by multi-layer convolution and pooling operations.Then a residual module was introduced to address the issue of gradient disappearance in the deep network.Secondly,the defect classification was carried out according to the extracted features.Finally,the key network parameters were optimized according to the leather data set.A data enhancing method was utilized to augment the dataset,which effectively avoided the over-fitting problem caused by insufficient data volume.The experimental results indicate that the proposed method can accomplish the classification according to the defect types and the classification accuracy can reach 92.34%.
作者
邓杰航
吴昌政
梁鸿津
顾国生
翁韶伟
DENG Jie-hang;WU Chang-zheng;LIANG Hong-jin;GU Guo-sheng;WENG Shao-wei(School of Computers,Guangdong University of Technolog,Guangzhou 510006,China;School of Computer Science and Software,Zhaoqing University,Zhaoqing 526061,China;School of Information Engineering,Guangdong University of Technology,Guangzhou 510006,China)
出处
《科学技术与工程》
北大核心
2020年第8期3143-3148,共6页
Science Technology and Engineering
基金
国家自然科学基金(61872095,61571139,61201393,61202267)
广东省信息安全技术重点实验室开放课题基金(2017B030314131)
广东省智能信息处理重点实验室、深圳市媒体信息内容安全重点实验室2018年开放基金课题(ML-2018-03)。
关键词
皮革图像识别
自动分类
参数优化
残差网络
数据增强
recognition of leather images
automatic classification
parameter-optimized
residual network
data augmentation
