Research And Application Of Garbage Classification System Based On Deep Learning

Garbage classification have great significance to garbage recycling and environmental protection.Since Shanghai implemented the regulations on garbage classification management in 2019,our country has achieved initial results on garbage classification project,the phenomenon of mixed waste is still existed.The reason is people are not clear enough about the rules of garbage classification.Therefore,developing a garbage classification.System can reduce the cost of our learning,but also can solve the garbage classification troubles in our daily life.Because of this situation,this paper set a rule that based on the Qingdao garbage classification standard,utilize the deep convolutional network to get a garbage image classification and recognition model,and develope a system for garbage classification on the mobile terminal based on this model.The main work of this paper is as follows.(1)In order to analyze the mobile terminal application software garbage advantages and disadvantages,the paper investigates the dynamics of their;Analyzed the characteristics of convolutional neural networks and the application of deep learning in the field of garbage classification.It focus on the structure of Inception in traditional CNN Google Net and mobile CNN Mobile Net.(2)This paper provides a garbage image classification model MobileNet-Inception based on CNN.We must consider the model's accuracy of the Mobile-based garbage classification.At the same time,we should ensure the model is lightweight,which means the amount of parameters cannot be too large,so it's necessary to consider both of them when designing the model.In order to ensure the model is lightweight,the model is based on Mobile Net,considering the large differences amount the garbage features,the depth separable convolutions which have different sizes are incorporated into the model,so as to obtain different scales of receptive fields and improve the accuracy of the model.(3)This paper establishes a garbage image classification data set.In order to know the garbages which are usually misclassified,this paper obtained the misclassified garbage catalog through questionnaire survey,according to that,by collecting the pictures of garbage on Internet.Arranging the exist garbage image datasets and taking pictures of the garbage around us,a self-made Garage Image data set is obtained.(4)The Android application of garbage image classification based on mobile incisionnet is designed and implemented.The system is designed with Spring MVC architecture and separation of client and service ends.The main function of the system is to realize the classification and identification of garbage by taking pictures of mobile phones.Other functions include registration and login,garbage inquiry,picture recognition,knowledge popularization,search logs,etc.