Waste Sorting Using Attention Mechanism Based on Deep Learning

The significance of waste sorting lies in achieving effective recycling of resources, reducing environmental pollution, and promoting sustainable development. Traditional waste sorting mainly relies on manual labor, but this method is inefficient, costly, and prone to misclassification issues. Introducing Convolutional neural networks (CNNs) and attention mechanisms can achieve automatic waste sorting, improve efficiency, reduce costs, and decrease reliance on manual labor. The attention mechanism is a mechanism that allows the model to focus on important information, which helps to enhance model performance. In waste sorting, introducing the attention mechanism enables the model to concentrate more on the key areas within waste images, thus improving the accuracy of classification. In this paper, we have embedded shuffle attention (SA) into ShuffleNetV1, and in a multi-model comparison, it achieved the best waste sorting accuracy, with an accuracy of 87.84%, recall of 85.75%, precision of 87.23%, kappa of 84.80%, and f1-score of 85.96%. This provides an innovative and feasible solution for the automation of waste sorting.