An image classification method based on separable convolution and attention mechanism

Abstract

The invention relates to an image classification method based on a separable convolution and attention mechanism, which is characterized by including the following steps: S1, constructing an originaldepth convolution neural network; 2, training an original deep convolutional neural network by using the train data set to obtain a trained deep convolutional neural network; S3, inputting the verification data set into the trained deep convolutional neural network to obtain a classification probability vector; 4, selecting that classification corresponding to the maximum probability value in theclassification probability vector as a test result of data preprocessing; S5, comparing the test result with the category label of the verification data set to obtain the accuracy of the final classification. The image classification method based on a separable convolution and attention mechanism provided by the invention can be used for improving the efficiency of feature extraction of the imageclassification model in deep learning, reducing the training parameter quantity of the model, and accelerating the model convergence.

Description

technical field [0001] The invention belongs to the field of computer vision, and in particular relates to an image classification method based on a separable convolution and attention mechanism. Background technique [0002] Image classification is an important fundamental problem in the field of computer vision and the basis of object recognition. With the development of Internet technologies such as online shopping and social networking, the number of digital images has also increased dramatically. Higher requirements are put forward for the analysis, processing and classification of these digital images, and high-efficiency and high-accuracy classification methods have become the key to solving such problems. [0003] There are two types of existing image classification methods: one is based on the calculation of joint distribution probability classification, such as Gaussian model, hidden Markov model, etc.; the other is based on the calculation of conditional probabil...

AI Technical Summary

Problems solved by technology

[0004] However, the existing methods are not sensitive to noise in the image and important information such as detail texture, color information, etc., resulting in a failure to achieve a good classification accuracy

Images