Research on Garbage Image Classification Based on Neural Network

With the advent of the global information era, the traditional waste image sorting algorithm can no longer meet the needs of modern urban waste sorting, and modern society has an increasingly strong desire for intelligent automation of waste image sorting technology. Therefore, the research and innovation of waste image sorting technology are essential for the development of the computer field and the modernization process of society. For garbage image classification, to solve the problem of garbage classification, a pre-trained ResNet34 model with migration parameters based on the ImageNet dataset is proposed in this paper to solve this problem. The model uses batch normalization to improve network stability and accelerate network convergence. At the same time, a lightweight CBAM attention mechanism module is added to enhance the focus on the main features of images and ignore irrelevant features to improve network performance further. The experimental results show that the performance of the improved model is considerably improved, and the performance results are better on the relevant garbage classification dataset, which has some practical value.