Application Of Image Classification Algorithm Based On Deep Learning In Garbage Classification

Nowadays,more and more attention has been paid to the garbage classification.Various ministries and commissions actively promote the relevant work.How to effectively carry out waste classification has become a valuable research topic.At present,China’s garbage classification mainly relies on manual sorting,which has low efficiency and high error rate,and also causes serious pollution to the environment.Therefore,an efficient and automatic garbage classification method needs to be proposed.Aiming at the problem of garbage image classification,referring to the current mainstream image classification algorithms,this paper proposes a garbage image classification algorithm based on ResNeXt,combined with attention mechanism module and dilated conv,and tests the algorithm in this paper on public data sets,and proves the feasibility of this algorithm.This paper mainly includes the following aspects:1.The research of this paper is based on the public data set provided by "the garbage classification challenge cup of Huawei cloud artificial intelligence competition" held by Huawei in 2019.For the samples in the data set,image preprocessing and data enhancement methods such as image standardization,image spatial geometric transformation,random erasure and Mixup are carried out,and the effectiveness of the above methods is compared and analyzed.2.In this paper,ResNeXt is used as the backbone network of classification network.Through the comparative experiments with other basic networks,it is proved that ResNeXt has a better effect on the classification problem in this paper.After that,the attention mechanism module is added to the backbone network,and comparative experiments show that the introduction of attention mechanism module can improve the accuracy of spam image classification results.3.In this paper,dilated conv is used to replace the general convolution in the backbone network topology.Experiments and analysis show that the use of dilated conv can effectively improve the network performance and greatly increase the accuracy of image classification.The classification algorithm proposed in this paper can accurately classify the samples in the test set of "Huawei cloud artificial intelligence competition garbage classification Challenge Cup".The classification accuracy rate of all samples reaches 95.9%,and the classification accuracy rate of each sub category is more than 90%,which can meet the actual needs of garbage image classification task.At the same time,this paper also gives the direction of further research on this algorithm.