Development method and system of small sample classification model based on graph convolutional neural network

Abstract

The invention belongs to the field of computer vision, and specifically relates to a method and system for developing a small-sample classification model based on a graph convolutional neural network, aiming to solve the problem of poor performance of existing models applied to new tasks and the need for a large number of labeled training samples The problem. The present invention includes: extracting a knowledge graph composed of an undirected graph to obtain knowledge information related to tasks; extracting experience information in the training process of the original model for old tasks; combining knowledge information and experience information to form a A new fusion graph; by establishing the connection between the new task category and the old task category and training the model, a classification model after cognitive development for the new task is obtained. In the process of task migration, the method of the present invention can realize fast and accurate visual migration without providing a large number of labeled samples for the new task, greatly improves the utilization rate of the model, and reduces the cost and time of training the model.

Description

technical field [0001] The invention belongs to the field of computer vision, and in particular relates to a method and system for developing a small sample classification model based on a graph convolutional neural network. Background technique [0002] With the continuous development of related technologies in the field of computer vision, the application scope of computer vision technology is becoming more and more extensive. With the gradual maturity of visual algorithms, the dependence of models on large-scale image samples has become an unavoidable problem in model training. For the fields where sample collection is more difficult, or even the fields that do not have large-scale samples, the collection of labeled samples has become a difficult problem restricting the development of computer vision technology. As the field of human cognition continues to expand, new task categories are gradually being discovered. However, the test task objects of most target recogniti...

AI Technical Summary

Problems solved by technology

[0005] In order to solve the above-mentioned problems in the prior art, that is, the performance of existing models is poor when applied to new tasks and a large number of labeled training samples are required, the present invention provides a small-sample classification model based on graph convolutional neural network. A developmental approach that includes:

Images