基于深度学习的有机番茄鉴别  

Organic tomatoes identification based on deep learning framework

作　　者：陈思维 李嘉璐 卢哲 颜文婧 CHEN Siwei;LI Jialu;LU Zhe;YAN Wenjing(School of Light Industry Science and Engineering,Beijing Technology and Business University,Beijing 100048,China;School of Computer and Artificial Intelligence,Beijing Technology and Business University,Beijing 100048,China)

机构地区：[1]北京工商大学轻工科学与工程学院,北京100048 [2]北京工商大学计算机与人工智能学院,北京100048

出　　处：《安徽农学通报》2024年第13期95-102,共8页Anhui Agricultural Science Bulletin

基　　金：北京市属高校教师队伍建设支持计划高水平科研创新团队项目(BPHR20220104)。

摘　　要：为满足消费市场上不断增长的有机番茄食用与鉴别需求,基于质谱检测数据,本文研究了一种有机番茄快速鉴别深度学习模型。首先,模型使用无监督降维方法对原始质谱检测数据进行降维,提取关键信息;其次,使用长短期记忆网络(Long short-term memory,LSTM)和Transformer网络提取序列信息特征;最后,利用反向传播(Back propagation,BP)神经网络构建分类器,实现面向有机及非有机番茄的精准识别。模型识别准确率在训练集上表现为98.437%,在测试集上表现为97.478%。结果表明,模型在有机及非有机番茄质谱快速检测识别任务上具有一定应用潜力,可部分满足有机番茄市场的发展需求,为有机番茄鉴别提供一定参考。In order to satisfy the growing demand for organic tomatoes consumption and identification in the consumer market,a deep learning model for rapid identification of organic tomatoes was researched based on mass spectrometry detection data.Firstly,the model used the unsupervised dimensionality reduction method to reduce the dimensionality of the original mass spectrometry detection data and extract key information.Secondly,long short-term memory(LSTM)and Transformer network were used to extract sequence information features.Finally,back propagation(BP)neural network was used to construct classifiers to achieve accurate recognition of organic and non-organic tomatoes.The recognition accuracy of the model was 98.437%on the training set and 97.478%on the test set.The results indicated that the model had potential for application in the rapid detection and identification of organic and nonorganic tomatoes mass spectrometry tasks,which could partly meet the development needs of the organic tomatoes market and provide references for the identification of organic tomatoes.

关 键 词：深度学习 有机番茄 降维 神经网络 长短期记忆网络 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术] S641.2[自动化与计算机技术—计算机科学与技术]