Image Recognition and Classification Method of Power Equipment Based on Convolutional Neural Network

A solution based on Convolutional Neural Network (CNN) is proposed to address the complex and diverse image recognition and classification challenges of power equipment. By constructing and training deep learning models, different types of power equipment images can be recognized and classified. This paper investigates the effect of different factors on CNN algorithms in power equipment image recognition and classification. Firstly, the effects of different data enhancement techniques are compared, where the average accuracy of the model using the scaling technique is 0.878. In an evaluation of the effect of different network depths on model performance, it is found that the 3-layer convolutional CNN algorithm performs better in terms of P-R curves, with precision decreasing from 0.90 to 0.70, and recall increasing from 0.60 to 0.80. In an experiment of the effect of different optimizer, the loss function of Adam's optimizer decreases rapidly from 0.8 to 0.36. In the final dataset size impact experiment, the larger the dataset size, the better the model performance. From the conclusion of the above data, the CNN algorithm cited in this paper effectively improves the accuracy and efficiency of power equipment image recognition and classification.