摘要
用微控制器实现人工智能是近年来嵌入式产品开发的热点和难点。本文讨论在MCU上部署深度网络模型,在MCU上进行推理的方法、技术和工具,分析部署MCU AI的流程、MCU深度网络模型支持库以及网络模型转换工具和优化方法,最后介绍知名MCU供应商发布的深度网络模型训练和部署工具。
Using microcontroller to realize artificial intelligence is a hot and difficult point in embedded product development recently.This paper discusses the methods,technologies and tools of deploying deep network model MCU,for reasoning on MCU and realizing MCU AI.The paper analyzes the basic implementation process of MCU AI,and the MCU deep network model supported libraries,neural network model conversion tools and optimization methods.Finally,this paper introduced some typical MCU deep network model training and deployment tools.
作者
李冬冬
林金龙
Li Dongdong;Lin Jinlong(Chifeng Construction Vocational and Technical School,Chifeng 024005,China;School of Software and Microelectronics,Peking University)
出处
《单片机与嵌入式系统应用》
2021年第12期6-11,共6页
Microcontrollers & Embedded Systems
