Garbage Recognition And Classification Based On Attention Mechanism And Convolutional Neural Network

At present,the material conditions in our country are increasing,but the accompanying garbage is also increasing,and it has become extremely impractical to sort waste manually.In recent years,the field of machine learning has developed rapidly and some models in the field of image classification are now available for use,and a variety of required data is easily available.In this context,it is a trend to leave the task of waste classification to computers,which saves resources and improves the efficiency of waste classification.Therefore,this paper will be based on convolutional neural networks for waste recognition and classification to provide more ideas for intelligent and automatic waste classification.The specific research content is as follows: firstly,for the field of waste classification studied in this paper,three models with high accuracy rate of image classification,respectively called Inception model,Res Net model and Dense Net model,at the same time,The data set of garbage sorting images required in this paper is constructed by crawler and manual shooting.We also compare the results of the three models to find out some problems of each of them and their advantages.Secondly,in order to improve the accuracy of the model,the attention mechanism is applied to the neural network model.Finally,the theoretical knowledge of target detection is applied to the field of waste classification in order to realize the multi-target classification task.The innovations of this paper mainly include:(1)Introducing the attention mechanism into the Res Net model and applying it to the field of garbage classification.(2)Apply the YOLOv7 to the field of garbage classification.Through the above study,we can obtain the following conclusions:(1)The dataset constructed in this paper has better performance on the Res Net model.(2)The inclusion of the attention module can significantly improve the accuracy of the dataset on the CNN.(3)The YOLOv7 target detection model can well solve the shortcomings of convolutional neural networks,that is to achieve the multi-target classification recognition task.