Face image clustering method and device for link prediction based on self-attention mechanism

Abstract

The invention discloses a face image clustering method and device for link prediction based on a self-attention mechanism, and the method comprises the steps: 1, selecting samples, supposing that the total number of the samples is N, and carrying out the feature extraction of the selected samples through a face recognition model; the method comprises the steps of 1, inputting a candidate enhancement feature set of an ith sample into a feature enhancement coding module based on context information for enhancement, 2, inputting the candidate enhancement feature set of the ith sample into a relation coding module with self-attention to obtain all possible link sets of the ith sample, and 4, combining the link sets of all the samples through a union-check set algorithm to obtain a candidate enhancement feature set of the ith sample. And a final clustering result is obtained. According to the method, a clustering task can be converted into a classification task through link prediction, and the accuracy of a clustering result can be improved; the effect of enhancing original node features is achieved by extracting and combining context information of part of neighbor nodes, and the negative influence of samples with low distinction degree is reduced.

Description

technical field [0001] The invention relates to the research fields of image recognition and image processing, in particular to a face image clustering method and device for link prediction based on a self-attention mechanism. Background technique [0002] At present, as the scale of face data sets becomes larger and larger, manual labeling requires a lot of manpower and material resources. Face clustering methods can greatly reduce the workload of data labeling. [0003] In the process of realizing the present invention, the inventor found that there are at least the following problems in the prior art: the existing face clustering methods generally make different assumptions directly on the input features, for example, the DB-SCAN algorithm requires the density of each cluster greater than a certain threshold. Various clustering algorithms using a single dimension, such as: clustering schemes based on traditional clustering algorithms, face clustering schemes based on adj...

AI Technical Summary

Problems solved by technology

[0003] In the process of realizing the present invention, the inventor found that there are at least the following problems in the prior art: the existing face clustering methods generally make different assumptions directly on the input features, for example, the DB-SCAN algorithm requires the density of each cluster greater than a certain threshold

Various clustering algorithms using a single dimension, such as: clustering schemes based on traditional clustering algorithms, face clustering schemes based on adjacency distance measurement, face clustering schemes based on hierarchical clustering, or several of them combination, etc., the accuracy of this type of clustering scheme is low

In addition, there are always samples with low discrimination, and the links between these sample features are usually difficult to predict directly

It is also difficult to improve the accuracy through traditional methods and recent deep learning methods