基于机器视觉的有色金属加工物料识别跟踪技术研究  

Research on Material Identification and Tracking Technology of Non-Ferrous Metal Processing Based on Machine Vision

作　　者：刘兴刚 刘为超[2] Liu Xinggang;Liu Weichao(China Nonferrous Metals Processing Technology Co.,Ltd.,Luoyang 471039,China;Luoyang Normal University,Luoyang 471934,China)

机构地区：[1]中色科技股份有限公司,河南洛阳471039 [2]洛阳师范学院,河南洛阳471934

出　　处：《有色金属加工》2022年第5期64-67,共4页Nonferrous Metals Processing

摘　　要：针对传统物料识别、跟踪技术无法实施在线跟踪的问题,提出一种基于深度学习的有色金属加工物料跟踪技术,该技术先采用Yolov3识别物料,然后采用DeepSort跟踪物料,最后通过二维和三维坐标转换获取现实物料库的定位信息。通过pycharm平台实验验证,该方案能够应用到实际应用中,满足物料识别和跟踪的准确性。In order to solve the problem that traditional material identification and tracking technology cannot implement online tracking, this paper proposes a material tracking technology based on deep learing for non-ferrous metal processing. The technology firstly uses Yolov3 to identify materials, then uses DeepSort to track materials, and finally obtains the location information of the real material library through 2D and 3D coordinate transformation. Pycharm experiment verified that the scheme could be applied to practical applications and meet the accuracy of material identification and tracking.

关 键 词：智慧工厂 物料识别 物料跟踪 Yolov3 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]