An Optimized Technique for Image Classification Using Deep Learning

Abstract

We propose a deep learning image classification model that aimed to serve as a framework and support for the recognition of big datasets of images. This paper explained the many forms of convolution neural networks and the fundamental steps involved in their use in image classification, starting with an analysis of the fundamental theory of neural networks. Secondly, an image classification deep learning model was proposed based on the improved convolution neural network structure, and noise reduction and parameter adjustment were carried out in the feature extraction process based on the current convolution neural network model. Finally, the deep learning model's structural optimisation was done to raise the model's classification efficiency and accuracy. Through studies, the association between the accuracy of many popular network models in image classification and the number of iterations was evaluated in order to confirm the effectiveness of the deep learning model proposed in this study in image classification. The model put forward in this paper was superior to other models in classification accuracy, according to the data. The training set and test set were used to evaluate and analyse the classification accuracy of the deep learning model before and after optimisation. The results demonstrated that, after the model recommended in this study had been fairly optimised, the accuracy of image classification had been substantially enhanced.