摘要
本文探讨工业场景下仪器仪表识别与分类技术。识别方面提取图像与数据特征,利用如CNN等多种算法。分类采用新的多模态融合方式与动态特征调整算法相结合,可根据不同工业场景实时调整特征权重。实验涵盖FF-DT、MM-FLF-SVM等多种模式,结果显示FF类平均处理时间短,MM类在准确率与稳定性上佳。实际应用需按工业场景需求选方法,实现高效识别分类。
This paper discusses the instrument recognition and classification technology in industrial scenarios.In terms of recognition,image and data features are extracted,and multiple algorithms such as CNN are used.For classification,a new multi-modal fusion method is combined with a dynamic feature adjustment algorithm,which can adjust feature weights in real time according to different industrial scenarios.The experiments cover multiple modes such as FF-DT and MM-FLF-SVM.The results show that the FF category has a short average processing time,and the MM category has excellent accuracy and stability.In practical applications,methods should be selected according to the requirements of industrial scenarios to achieve efficient recognition and classification.
作者
田洪伟
撒兴才
代伍年
曹汝庆
Tian Hongwei;Sa Xingcai;Dai Wunian;Cao Ruqing(Beijing Pulong Technology Co.,Ltd.,Xiongan,China;Construction Headquarters of a Project,Logistics Support Department,Central Military Commission,Beijing,China;China Railway 14th Bureau Group Electrification Engineering Co.,Ltd.,Jinan,China)
出处
《科学技术创新》
2025年第7期1-4,共4页
Scientific and Technological Innovation
关键词
工业场景
人工智能
仪器仪表
识别分类
industrial scenario
artificial intelligence
instrument
recognition and classification
