摘要
随着制药行业的快速发展,药品质量管理愈发重要,药粒的检测与位置识别成为药品生产和配送中的关键环节。传统的药粒检测方法存在速度慢、精度低等问题,难以满足现代化生产需求。文章提出了一种基于改进YOLOv5的药粒检测与位置识别方法,该方法通过引入注意力机制、优化损失函数和采用数据增强策略,有效提升了模型在复杂背景下的鲁棒性和对药粒目标的识别精度。实验结果表明,改进后的YOLOv5模型在药粒检测任务中,其精度和召回率相比传统方法有显著提高,且在实时性和鲁棒性方面表现出色,这为后续的药品包装和自动化操作提供了有力支持。
With the rapid development of the pharmaceutical industry,drug quality management has become increasingly important,and the detection and position recognition of drug pills have emerged as the key links in drug production and distribution.Traditional drug pill detection methods suffer from problems such as slow speed and low accuracy,making them difficult to fulfill the requirement of modern production.This paper proposes a method for drug pill detection and position recognition based on improved YOLOv5.By introducing an attention mechanism,optimizing the loss function as well as adopting data augmentation strategies,the robustness of the model in complex backgrounds and the recognition accuracy for drug pill targets are effectively improved.Experimental results show that the improved YOLOv5 model achieves significantly higher accuracy and recall rate compared to the traditional methods in drug pill detection tasks,and performs excellently in terms of real-time performance and robustness,thus providing strong support for subsequent drug packaging and automated operations.
作者
刘子淳
郎庆阳
许鹏
LIU Zichun;LANG Qingyang;XU Peng(School of Mechanical Engineering,Liaoning Institute of Science and Technology,Benxi Liaoning 117004,China)
出处
《辽宁科技学院学报》
2025年第5期38-41,50,共5页
Journal of Liaoning Institute of Science and Technology
基金
辽宁省教育科学“十四五”规划2024年度课题“地方应用型高校机械专业学生创新实践能力培养模式研究”(JG24DB290)
2022年度辽宁科技学院本科教学改革研究立项项目“基于实验室建设的机械专业学生创新实践能力培养模式的研究与实践”
2024年辽宁科技学院大学生创新创业训练计划项目“智能家庭用药辅助装置”(202411430025).
