基于YOLOv5的药瓶识别定位系统设计  

Design of a Medicine Bottle Recognition and Localization System Based on YOLOv5

作　　者：孔凡国[1] 仇展明 王鑫 袁功兴 陈靖轩 KONG Fanguo;QIU Zhanming;WANG Xin;YUAN Gongxing;CHEN Jingxuan(School of Mechanical and Automation Engineering,Wuyi University,Jiangmen 529020,China)

机构地区：[1]五邑大学机械与自动化工程学院,广东江门529020

出　　处：《机械工程与自动化》2025年第1期188-189,192,共3页Mechanical Engineering & Automation

摘　　要：研究设计了一种药瓶识别定位系统,实现了对图像中药瓶的识别与定位。该系统使用奥比中光Astra Pro Plus相机进行图像采集,利用YOLOv5深度学习模型检测画面中的药瓶,并计算药瓶中心坐标,输出给机器人使其完成抓取任务。通过实验验证,该系统在测试集上的平均精度均值为92.72%,表明该系统在药瓶检测与定位上具有可行性和较高的准确率。This study presents a medicine bottle recognition and localization system,achieving identification and positioning of medicine bottles in images.The system utilizes the ORBBEC Astra Pro Plus camera for image acquisition and employs the YOLOv5 deep learning model to detect medicine bottles in the captured images.The system calculates the central coordinates of the medicine bottles and outputs them to the robot to complete the grasping task.The system achieves an average precision of 92.72%on the test set.Through experimental analysis,the system demonstrates feasibility and a certain level of accuracy in medicine bottle detection and localization.

关 键 词：药瓶 识别定位 深度学习 准确率 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]