基于深度学习的垃圾分类系统  

Garbage classification System Based on Deep Learning

作　　者：李丕兵 孙仁诚[1] LI Pibing;SUN Rencheng(Qingdao University,Qingdao Shandong 266071,China)

机构地区：[1]青岛大学,山东青岛266071

出　　处：《信息与电脑》2021年第4期43-45,共3页Information & Computer

摘　　要：垃圾分类是近几年提出的概念,各地政府都在大力推行,但解决垃圾分类问题是一个难题,因为各地政府的政策都不一样,加上科普力度低,很多人只能凭借常识来区分垃圾,但面对难以区分的垃圾时并不能正确判断。基于此,笔者针对青岛市垃圾分类政策并结合对居民问卷调查设计了一款垃圾分类系统。在常用CNN模型中,inception-v3模型与其他模型相比具有更高的精度,实验表明基于inception-v3的垃圾分类模型准确率在90%以上,能够在一定程度上解决垃圾分类困难的问题。Garbage classification is a concept that has been put forward in recent years. Local governments have been vigorously promoting it, but solving the problem of garbage classification is a difficult problem. Because the policies of local governments are different, and the intensity of science popularization is low, many people can only distinguish garbage based on common sense., But can’t judge correctly when faced with indistinguishable garbage. Based on this, the author designed a garbage classification system based on Qingdao’s garbage classification policy and combined with the residents’ questionnaire survey. Among the commonly used CNN models, the inception-v3 model has higher accuracy than other models. Experiments show that the accuracy of the garbage classification model based on inception-v3 is above 90%, which can solve the problem of garbage classification to a certain extent.

关 键 词：深度学习 垃圾分类 inception-v3模型 迁移学习 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TP183[自动化与计算机技术—计算机科学与技术]