Detection of Contraband in Milk Powder Cans by Using Stacked Auto-Encoders Combination with Support Vector Machine

2nd International Symposium on Resource Exploration and Environmental Science

生态环境科学

Zhu, Yuping^1 ; Wang, Lei^2 ; Zhang, Wei^2

Library, Zaozhuang University, Shandong, China^1

Information Science and Engineer Department, Zaozhuang University, Shandong, China^2

关键词: 5-fold cross validation method;    Auto encoders;    Contraband detection;    Detection device;    Grid methods;    Material resources;    Model-based OPC;    Railway stations;   

Others  :  https://iopscience.iop.org/article/10.1088/1755-1315/170/3/032114/pdf
DOI  :  10.1088/1755-1315/170/3/032114

学科分类：环境科学（综合）

来源: IOP

【 摘 要 】

The carrying of contraband has brought increasingly serious harm to people's lives. At present, detection devices used in important places such as customs, airports and railway stations can not automatically identify contraband, and the final identification is entirely done by hand. Therefore, all countries have devoted a great deal of manpower and material resources to studying and developing more effective contraband detection technologies. In this article, we propose a model based on the stacked auto-encoders (SAE) method to detect contraband in milk powder cans. Firstly, we construct a representative of the majority of the reality of the milk CT image data set, secondly, we use the SAE method to extract the features, and finally use the support vector machine (SVM) classifier to determine whether the contracted product is carried in the milk powder cans. In order to prevent the data from over fitting, in the experiment we used the 5-fold cross-validation method. In addition, we also use the grid method to adjust the parameters of SVM. The excellent experimental results show that the model we proposed has a good effect on the detection of carrying contraband in milk powder cans.