{"title":"一种危险设备检测的深度学习方法","authors":"Jianxin Yuan, Chengan Guo","doi":"10.1109/ICIST.2018.8426165","DOIUrl":null,"url":null,"abstract":"Effective detection of concealed dangerous equipment is a critical need to protect people’ security in crowd public situations. Terahertz (THz) technology is ideally suited for such an application since it is able to see through clothing and packages, and, in addition, THz photons have lower energy than infrared and do not show the ionizing properties of X-ray radiation. There are two key technologies involved in this application: one is to develop THz imaging hardware and the other is to develop corresponding machine vision algorithms. In this paper we address to the latter and develop a deep learning-based method for detection and recognition of the dangerous equipment in THz images. The detection method is implemented with a two-stage classifier, in which the first-stage classifier is for detecting the direct visible dangerous equipment in natural light images, and the second-stage classifier is for detecting the concealed dangerous objects in THz images. In the detection system, when an input image is classified as a natural image, it is directly processed to give final classification result by the first-stage classifier. While the input image is classified as a THz image, it is sent to the second-stage classifier for finer processing and classification. Preliminary experiments conducted in the work show that the proposed method can give satisfactory performance in detection/recognition of dangerous equipment both in nature and THz images.","PeriodicalId":331555,"journal":{"name":"2018 Eighth International Conference on Information Science and Technology (ICIST)","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2018-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"16","resultStr":"{\"title\":\"A Deep Learning Method for Detection of Dangerous Equipment\",\"authors\":\"Jianxin Yuan, Chengan Guo\",\"doi\":\"10.1109/ICIST.2018.8426165\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Effective detection of concealed dangerous equipment is a critical need to protect people’ security in crowd public situations. Terahertz (THz) technology is ideally suited for such an application since it is able to see through clothing and packages, and, in addition, THz photons have lower energy than infrared and do not show the ionizing properties of X-ray radiation. There are two key technologies involved in this application: one is to develop THz imaging hardware and the other is to develop corresponding machine vision algorithms. In this paper we address to the latter and develop a deep learning-based method for detection and recognition of the dangerous equipment in THz images. The detection method is implemented with a two-stage classifier, in which the first-stage classifier is for detecting the direct visible dangerous equipment in natural light images, and the second-stage classifier is for detecting the concealed dangerous objects in THz images. In the detection system, when an input image is classified as a natural image, it is directly processed to give final classification result by the first-stage classifier. While the input image is classified as a THz image, it is sent to the second-stage classifier for finer processing and classification. Preliminary experiments conducted in the work show that the proposed method can give satisfactory performance in detection/recognition of dangerous equipment both in nature and THz images.\",\"PeriodicalId\":331555,\"journal\":{\"name\":\"2018 Eighth International Conference on Information Science and Technology (ICIST)\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2018-06-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"16\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2018 Eighth International Conference on Information Science and Technology (ICIST)\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/ICIST.2018.8426165\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2018 Eighth International Conference on Information Science and Technology (ICIST)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/ICIST.2018.8426165","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
A Deep Learning Method for Detection of Dangerous Equipment
Effective detection of concealed dangerous equipment is a critical need to protect people’ security in crowd public situations. Terahertz (THz) technology is ideally suited for such an application since it is able to see through clothing and packages, and, in addition, THz photons have lower energy than infrared and do not show the ionizing properties of X-ray radiation. There are two key technologies involved in this application: one is to develop THz imaging hardware and the other is to develop corresponding machine vision algorithms. In this paper we address to the latter and develop a deep learning-based method for detection and recognition of the dangerous equipment in THz images. The detection method is implemented with a two-stage classifier, in which the first-stage classifier is for detecting the direct visible dangerous equipment in natural light images, and the second-stage classifier is for detecting the concealed dangerous objects in THz images. In the detection system, when an input image is classified as a natural image, it is directly processed to give final classification result by the first-stage classifier. While the input image is classified as a THz image, it is sent to the second-stage classifier for finer processing and classification. Preliminary experiments conducted in the work show that the proposed method can give satisfactory performance in detection/recognition of dangerous equipment both in nature and THz images.