Transformer Image Recognition System Based on Deep Learning

Due to the variety of transformer models and the complexity of the image, the traditional method based on machine learning for the artificial design features cannot be large-scale and accurate identification. This paper proposes a transformer image recognition system based on deep learning, which can acquire the target results directly by “end-to-end” learning from the original acquired images. In order to realize the accurate classification of transformer images, this paper selects convolutional neural network algorithm to apply to transformer image recognition, constructing a transformer image recognition system based on convolutional neural network by designing the network structure and classifier of convolutional neural network algorithm. By constructing training set and testing set of the transformer image, the model is trained and its performances are tested. It is shown in the experimental results that compared with BP neural network model and SVM model, convolutional neural network has better effects, reaching 9.17% of the identification error, and realizing the accurate distinction of three types of transformers.