Aiming at the traditional image classification model based on deep learning,an image classification method based on RepVGG is proposed due to the over fitting problem caused by the increase of network layer during the training process.This paper is optimized based on the RepVGG model,and the residual attention mechanism is used in the optimized model to enhance the extraction of image features by the network,and the full convolutional layer is used instead of the full connection to improve the processing ability of the model on image feature informa-tion.Experimental results show that the proposed method has high accuracy,and its highest ac-curacy is 96.3％,which proves the effectiveness of the optimization algorithm.
万方数据
维普
