Ethic group identification method based on integrated convolution neural network

Abstract

The invention discloses an ethic group identification method based on an integrated convolution neural network. The method comprises the following steps: S1, selecting M convolution neural network classifiers obtained through training by a use of a training set in an ethic group face database as basic classifiers; S2, obtaining face pictures to be detected; and S3, during a test, respectively inputting the face pictures to be detected into the M basic classifiers obtained in the first step, and then fusing ethic group types output by the M basic classifiers so as to obtain a final ethic group type. According to the method, an ethic group is identified through integration of the M convolution neural network classifiers, and compared to a conventional mode of identifying the ethic group by use of a single convolution neural network classifier, the method provided by the invention has the advantages of being high in ethic group identification accuracy, being capable of identifying multiple ethic groups and being wide in application.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to an ethnic identification method based on an integrated convolutional neural network. Background technique [0002] Facial ethnic recognition refers to the analysis of facial images, so that the computer can determine which ethnicity the face belongs to. Since there are 56 different ethnic groups in our country, and each ethnic group has its own unique ethnic characteristics, culture and customs, ethnic identification is of great significance. For example, in the catering industry, it can help businesses automatically determine the ethnic category of customers. To recommend a suitable diet. Moreover, each ethnic group has different cultural customs and entertainment habits. Once the ethnic category of the customer is identified, the merchant can recommend corresponding entertainment programs or commodities. In the design of the human-computer interaction interface of th...

AI Technical Summary

Problems solved by technology

Scholars at home and abroad mostly use these classification algorithms when solving the problem of face ethnic classification, but these classification algorithms greatly depend on the expression of the features of the face, and the feature extraction methods currently used are artificially designed and rely on the experience of artificial experts and repeated experiments, not only the workload is heavy, but also it is difficult to find an optimal expression of ethnic characteristics

In addition, at present, ethnic identification is mainly foreign ethnic identification, and there is still little research in this area in China. In particular, there is no ethnic identification standard database available for research in China, which makes the progress of ethnic identification slow. This also shows that the face The difficulty of ethnic identification, the accuracy of identification needs to be improved urgently

Images