Model training method, kinship classification method, retrieval method and related devices

Abstract

The invention provides a model training method, a kinship classification method, a retrieval method and related devices. The model training method comprises the steps of obtaining a training set; obtaining a first feature vector corresponding to the first training image and a second feature vector corresponding to the second training image; acquiring a third feature vector corresponding to the first training image and a fourth feature vector corresponding to the second training image; inputting the feature vector into a splicing layer to obtain splicing information; inputting the splicing information into a full connection module to obtain prediction information; and training a preset deep neural network by using the prediction information and annotation information to obtain a kinship classification model. The prediction information is obtained by splicing the output results of the two feature extraction modules, so that the precision of the obtained prediction information is higher, and the precision is correspondingly higher when the kinship classification model trained by using the prediction information is used for kinship classification or retrieval.

Description

Technical field [0001] This application relates to deep learning and image processing technology, in particular to model training method, kinship classification, retrieval method and related devices. Background technique [0002] With the development of the baby home and other public activities, the community for relief operations lost children is increasing, the identity of these missing children has become a big problem determination. Very urgent need for a convenient way to save these broken homes, hopes of finding their biological parents of missing children. [0003] Determine the biological relationship of theory based on Mendelian genetics, judging by identifying DNA sequences of both parties whether related by blood. However, due to the relationship between time and space, and can not do anywhere D NA test, but due to genetic reasons, most people's appearance with their blood are similar, therefore, to be verified by kinship feasibility photo becomes high, and to verify t...

AI Technical Summary

Problems solved by technology

[0008] The purpose of this application is to provide a model training method, a kinship classification method, a retrieval method and related devices, so as to solve the problem that the accuracy of parent-child relationship judgment using images is still low in the prior art

Images