Human age estimation method based on self-adaptation sign distribution

Abstract

The invention discloses a human age estimation method based on self-adaptation sign distribution. According to the method, through a self-adaptation training method combined with the application of an age sign distribution model algorithm, the obtained facial image feature vectors and initial age sign distribution are used as inputs, the KL divergence between the input initial age sign distribution and model prediction sign distribution is minimized by applying the age sign distribution model algorithm, and therefore the prediction age sign distribution can be obtained. Then, the prediction age sign distribution of the same age can be used for studying the age sign distribution of the corresponding age, circulation is carried out until training convergence is achieved or the set largest number of times is reached, and then the final prediction model can be obtained. Facial image features to be estimated are input into the final prediction model, and then the prediction age can be output. The age estimation accuracy of the system can reach the level similar to that of the human.

Description

technical field [0001] The present invention relates to the field of automatic estimation method of human age by computer. Background technique [0002] The application of automatic age estimation based on face images is becoming more and more widespread, mainly including the following aspects: (1) Age-based human-computer interaction system: on the basis of ordinary human-computer interaction systems, the human age automatic estimation algorithm is introduced, according to the user's (2) age-based access control system: used to prevent minors from accessing inappropriate web pages or content, purchasing tobacco and alcohol products on vending machines, entering bars, etc. Suitable places, etc.; (3) E-commerce: Estimate the approximate age of customers based on images, etc., and adopt different marketing strategies for customers of different age groups; (4) Criminal investigation: Based on image data of criminal suspects captured by video surveillance, etc. Determine the ap...

AI Technical Summary

Problems solved by technology

If you choose that all ages correspond to the same variance, that is, all samples correspond to the same Gaussian distribution, it is inconsistent with the fact that human age changes, because people's faces change faster and larger from infancy to adulthood, and from adulthood to old age The change is relatively slow and small, corresponding to the Gaussian distribution, the variance in the low age group should be relatively small and the variance in the high age group should be relatively large; on the other hand, if you choose different ages to correspond to different variances, that is, different ages The samples correspond to different Gaussian distributions, so the variances corresponding to different ages are difficult to set reasonably, lacking convincingness and credibility

Images