APP下载

基于TensorFlow Lite的雏鸡性别鉴别App设计

2021-08-18韩闰凯杨晶晶翟永辉顾志豪刘家兴郭甲倩

电脑知识与技术 2021年18期
关键词:机器学习

韩闰凯 杨晶晶 翟永辉 顾志豪 刘家兴 郭甲倩

摘要:近些年随着机器学习领域的不断发展,在诸多领域都展现出其独特的优势。目前在养鸡业中,雏鸡性别鉴别的方式主要为人工鉴别,鉴别过程需要耗费大量人力物力,且需要鉴别人员拥有丰富的鉴别经验。该文结合机器学习在图像识别方向的发展,将图像识别技术应用于雏鸡性别鉴别,从而解决养鸡行业中雏鸡性别鉴别问题。使用卷积神经网络构建雏鸡性别鉴别模型,并在高识别率的情况下对模型进行压缩,将其转换为Tensorflow Lite模型后,部署到Android App以实现自动鉴别雏鸡公母。同时,在Android App实现了鸡的品种查询、疾病查询等功能。该软件可以降低相关工作人员工作负担,提高鉴别雏鸡公母的效率,有重要的应用价值。

关键词:雏鸡识别;机器学习;养殖;Android;Tensorflow Lite

中图分类号:TP18      文献标识码:A

文章编号:1009-3044(2021)18-0082-04

开放科学(资源服务)标识码(OSID):

App Design for Identification of Chicken Sex Based on TensorFlow Lite

HAN Run-kai,YANG Jing-jing, ZHAI Yong-hui, GU Zhi-hao, LIU Jia-xing, GUO Jia-qian

(School of Information Science and Engineering, Hebei North University, Zhangjiakou 075000, China)

Abstract:Machine learning,with its continuous development in recent years in fields, its distinctive advance has appeared.  At present in chicken culture industry, the chicken sex is identified manually as the main, which costs much in man power and material, and the personnel is required experienced. In the paper,  development of machine learning in image identification is combined with the use of technology of image identification into chicken sex identification so as to solve the problem mentioned in chicken industry: apply Convolutional Neural Network(CNN) to construct chicken sex identification model, and compress the model in condition of high identification rate; transfer the compressed model to TensorFlow Lite model, then deploy the model to Android App for implementation of automatic identification of chicken sex and functions of enquiry of both chicken variety and disease, etc.  This software may be used to reduce relevant man power, improve the efficiency of identical rate of chicken sex, which is of important applicable value.

Key words: identification 0f chicken sex; machine learning; culture; android; TensorFlow Lite

1 背景

1.1 开发背景

随着生活水平的提高,人们对鸡肉及其副产品的需求量日益增加,导致养鸡业迅速发展。而初生雏鸡是较脆弱,饲养时需要较高的成本,同时由于温度、湿度、通风、光照等因素影响 [1],初生雏鸡拥有较高的死亡率。而且在养殖过程中,涉及的雏鸡性别识别问题[2],主要为人工操作,这在雏鸡养殖过程中消耗了大量的时间与人力。

现实生活中养鸡场无论是商品肉鸡或是商品蛋鸡养殖,都需要把不同性别的雏鸡分开养殖或按一定比例混合饲养[3],从而提高初生雏鸡的存活率和养殖的经济效益。……

登录APP查看全文

猜你喜欢

机器学习
基于词典与机器学习的中文微博情感分析
基于网络搜索数据的平遥旅游客流量预测分析
前缀字母为特征在维吾尔语文本情感分类中的研究
下一代广播电视网中“人工智能”的应用
基于支持向量机的金融数据分析研究
基于Spark的大数据计算模型
基于朴素贝叶斯算法的垃圾短信智能识别系统
极限学习机在图像分割中的应用
一种基于迁移极速学习机的人体行为识别模型