基于PaddlePaddle深度学习的野外露营安全装置  

A Field Camping Safety Device Based on PaddlePaddle Deep Learning

作　　者：刘映群 黄嘉源 何卓恒 李创铭 陈日源 LIU Yingqun;HUANG Jiayuan;HE Zhuoheng;LI Chuangming;CHEN Riyuan(Guangdong Open University(Guangdong Polytechnic Institute),Guangzhou 510091,China;School of Intelligent Manufacturing and Electrical Engineering,Guangzhou Institute of Science and Technology,Guangzhou 510540,China;School of Electrical and Computer Engineering,Nanfang College Guangzhou,Guangzhou 510970,China;Zhongshan Camry Electronic Company Limited,Zhongshan 528400,China)

机构地区：[1]广东开放大学(广东理工职业学院),广东广州510091 [2]广州理工学院智能制造与电气工程学院,广东广州510540 [3]广州南方学院电气与计算机工程学院,广东广州510970 [4]中山佳维电子有限公司,广东中山528400

出　　处：《现代信息科技》2024年第7期113-117,共5页Modern Information Technology

基　　金：2023年广东省科技创新战略专项资金(“攀登计划”专项资金)(pdjh2023b0845);2022年广东省大学生创新创业训练计划(S202213919003)。

摘　　要：针对近年野外露营存在的环境及安全隐患等问题,设计一种基于PaddlePaddle深度学习的野外露营安全装置。首先采用卷积神经网络训练模型,然后通过图像识别分类技术进行识别。文中以动物分类识别为例,通过对不同种类动物图片进行训练来改进模型,并对模型识别及分类的准确性进行验证,最终精度为0.96以上。经测试,采用的模型在动物图像分类中有较好效果,能满足课题设计要求。Aiming at the problems of environment and safety hazards in field camping in recent years,a field camping safety device based on PaddlePaddle Deep Learning(DL)is designed.Firstly,a Convolutional Neural Networks is used to train the model,and then image recognition and classification techniques are used for recognition.Taking animal classification and recognition as an example in the paper,the model is improved by training images of different types of animals,and the accuracy of model recognition and classification is verified.The final accuracy is above 0.96.After testing,the model used has shown good performance in animal image classification and can meet the requirements of the project design.

关 键 词：PaddlePaddle 卷积神经网络 图像识别分类 野外露营安全装置 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术] TP181[自动化与计算机技术—计算机科学与技术]