Image classification method based on improved residual network

Abstract

The invention belongs to the technical field of image classification, and particularly relates to an image classification method based on an improved residual network, which comprises the following steps: S1, acquiring a common data set, and completing the construction of a data set required by model training; s2, unifying the size of the data set, performing normalization, data division, denoising and other image enhancement methods, and performing random clipping and random erasing for data expansion to improve the model recognition accuracy; s3, using a hierarchical segmentation module andan attention mechanism to improve a residual error network model so as to improve the model identification performance and accelerate the operation time; and S4, when the loss function of the model does not drop any more and the evaluation index is optimal and tends to be stable, storing the model. According to the method, the problems that an existing network model is poor in image recognition effect and long in running time are solved, the ResNet50 model is improved by using a hierarchical segmentation module and an attention mechanism to be used for image recognition to promote stronger features of network learning, and the model recognition accuracy and the resource utilization rate are improved.

Description

technical field [0001] The invention belongs to the technical field of image classification, and in particular relates to an image classification method based on an improved residual network. Background technique [0002] Image classification, as the name suggests, is to label the input image with a fixed category. This is one of the central problems in the field of computer vision, and although it sounds simple, image classification has a large number of different practical applications. Image classification is one of the fundamental research topics in the field of artificial intelligence. [0003] In image recognition, the computing speed and computer resource consumption of the model are getting more and more attention, but because most of the current convolutional neural networks improve the recognition accuracy by designing deeper and wider network models, resulting in the increase of hyperparameters, the architecture Increasingly complex, making the model take an ext...

AI Technical Summary

Problems solved by technology

[0006] In order to overcome the deficiencies in the prior art, the present invention provides an image classification method based on the improved residual network, which can solve the problem that the existing network model has a poor effect on image recognition, the running time is longer, and the existing ResNet50 deep learning model has high computational complexity and long training period

Images