Multi-view image classification method and system, computer equipment and storage medium

Abstract

The invention provides a multi-view image classification method and system, computer equipment and a storage medium. The method comprises the steps of obtaining a multi-view image classification dataset; constructing a multi-view classification model based on a graph convolutional neural network according to the classification data set; wherein the multi-view classification model comprises an input layer, a spectrogram convolution layer, a batch regularization layer, a graph pooling layer, a full connection layer and a softmax function multi-classification output layer; and performing classification training on the multi-view classification model according to the classification data set to obtain a classification result. According to the embodiment of the invention, on the application ofmulti-view image classification, the processing of any data structure is supported, the universality of the model is improved, the low dimensionality of training data is ensured, the storage space andcomputing resources are reduced, and the convergence rate and classification precision of the model are improved under the condition that pre-training is not needed.

Description

technical field [0001] The present invention relates to the technical field of graph deep learning, in particular to a graph convolutional neural network-based multi-view image classification method, system, computer equipment and storage medium. Background technique [0002] A multi-view image is an image group composed of different perspective images of the same target object that can describe the target object more vividly, and the perspective that best represents the multi-view image target in the image group is usually called the optimal perspective. Multi-view images are widely used because they represent the target object more vividly than traditional single-view images, such as online display of products on shopping platforms, and naturally become the research object that scholars are keen on. [0003] Currently, multi-view descriptors can be divided into four categories: 2D image-based multi-view descriptors, voxel-based multi-view descriptors, 3D point cloud-based ...

AI Technical Summary

Problems solved by technology

However, these four multi-view descriptors have their own limitations. Among them, the multi-view descriptor based on two-dimensional images uses the classic convolutional neural network technology to fuse the two-dimensional features of different view images under the target image. Although the descriptor has an advantage in terms of time complexity, the classification accuracy needs to be further improved; voxel-based multi-view descriptors directly learn the features of 3D multi-view images and extract features through 3D convolution, but this method has high-dimensional input data; the multi-view descriptor based on 3D point cloud can directly process unordered 3D point cloud, but it is difficult to extract the local features and unstructured constraints of the point cloud; the multi-view descriptor based on graph convolutional neural network can handle arbitrary graph structure data, and can better describe the local characteristics of the data, but the classification accuracy needs to be further improved

Images