摘要
微生物识别对于水质检测及污水处理领域具有重要意义,传统方法效率低下,需要大型仪器、人工干预.针对上述问题,本文提出一种嵌入式平台下微生物数字全息显微图像分类系统.本文使用数字全息显微镜采集微生物图像,引入卷积神经网络进行分类计算,利用Tengine架构在嵌入式平台部署神经网络算法,构建实现了GoogLeNet、AlexNet、VGG16Net等模型,实现在Whoi微生物数据集上的分类检测.使用精简的GoogLeNet-Lite网络在国产嵌入式平台RK3399达到94.15%的准确率,以及12.7fps的运算速度.实验结果说明,采用卷积神经网络,在嵌入式平台RK3399上进行数字全息图像分类,既满足了快速检测的要求,同时也解决了系统体积问题,验证了本方法的有效性.
Microbial identification is of great significance in the field of water quality detection and sewage treatment. Traditional methods are inefficient and require large-scale instruments and manual intervention. In view of the above problems,this paper proposes a microbe digital holographic microscopic image classification system on an embedded platform. In this paper,a digital holographic microscope is used to collect microbial images,and a convolutional neural network is introduced for classification calculation. Tengine architecture is used to deploy neural network algorithms on the embedded platform. M odels such as GoogLeNet,AlexNet,VGG16 Net and other models are constructed and implemented to achieve classification on the Whoi microbial dataset. It achieves an accuracy of94. 15% and an operation speed of 12. 7 fps in the domestic embedded platform RK3399 by using a simplified GoogLeNet-Lite network. The experimental results show that the digital holographic image classification on the embedded platform RK3399 by using a convolutional neural network not only satisfies the requirements of rapid detection,but also solves the system volume problems,w hich verifies the effectiveness of this method.
作者
陈朋
戴陈统
宫平
王海霞
CHEN Peng;DAI Chen-tong;GONG Ping;WANG Hai-xia(College of Computer Science and Technology,Zhejiang University of Technology,Hangzhou 310023,China;College of Information Engineering,Zhejiang University of Technology,Hangzhou 310023,China)
出处
《小型微型计算机系统》
CSCD
北大核心
2020年第12期2595-2600,共6页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(U1909203)资助
浙江省重点研发计划项目(2019C01007)资助
浙江省属高校基本科研业务费专项资金项目(RF-C2019001)资助。
关键词
嵌入式系统
数字全息显微
卷积神经网络
微生物分类
embedded systems
digital holographic microscopy
convolutional neural networks
microorganism classification
作者简介
陈朋,男,1981年生,博士,教授,博士生导师,研究方向为模式识别、嵌入式系统设计;戴陈统,男,1993年生,硕士研究生,研究方向为模式识别、嵌入式系统设计;宫平,男,1995年生,硕士研究生,研究方向为模式识别、嵌入式系统设计;王海霞,女,1983年生,博士,副教授,研究方向为图像处理与模式识别.
