Face clustering method and device, and storage medium

Abstract

The invention discloses a face clustering method and device and a storage medium. The face clustering method comprises the following steps: performing query on a face feature KD tree according to the feature vectors to be clustered to obtain K neighbor feature vectors and the feature distances between the K neighbor feature vectors and the feature vectors to be clustered; sorting the K feature distances from small to large, selecting the corresponding M neighbor feature vectors as the close distance feature vectors, and calculating the symmetry distance between the close distance feature vectors and the feature vectors to be clustered; and associating the feature vectors to be clustered with the class corresponding to the close distance feature vectors if the symmetry distance is less than the clustering threshold. According to the invention, the clustered feature vectors are sorted by means of the KD tree, and the neighbor feature vectors are searched by means of the K neighbor algorithm, therefore, the search efficiency is high, the speed is fast, and the clustering range can be reduced quickly; the similarity between the feature vectors to be clustered and the close distance feature vectors is evaluated by means of the symmetry distance, and the feature vectors to be clustered are divided into the most similar classes, therefore, the accuracy of the clustering is ensured.

Description

technical field [0001] The present invention relates to video monitoring technology, in particular to a face clustering method, device and storage medium. Background technique [0002] In public places such as subways, stations, airports, and customs, under normal circumstances, a person will only appear once in a period of time. If a person appears many times, then this person's behavior will be worthy of the public security department and the public place management department. Focus. For example, people who appear many times at the ticket office of a station may be people looking for an opportunity to steal other people's property or scalpers who resell tickets; people who come and go repeatedly in the customs channel are more likely to be suspected of smuggling. For public places with a large flow of people, the number of abnormal occurrences is often associated with abnormal motivations. Locking people with abnormal occurrences in a specific place is of great significa...

AI Technical Summary

Problems solved by technology

[0006] In order to overcome the deficiencies of the prior art, one of the objectives of the present invention is to provide a face clustering method, which can solve the problem of low accuracy and computational complexity of existing face recognition schemes based on deep learning when the flow of people is large. Explosive problem

[0007] The second object of the present invention is to provide a face clustering device, which can solve the problems of low accuracy and sudden increase in the amount of computation in the existing deep learning-based face recognition scheme when the flow of people is large

[0008] The third object of the present invention is to provide a storage medium, which stores a computer program, which can solve the problems of low accuracy and sudden increase in the amount of computation in the existing face recognition scheme based on deep learning when the flow of people is large

Images