A method and device for object recognition based on dynamic graph

Abstract

The invention discloses an object recognition method and device based on a dynamic graph. The method includes: obtaining first recognition information, the first recognition information includes an object to be recognized and first interaction information corresponding to the object to be recognized; Divide the first identification information to obtain a plurality of second identification information, the second identification information includes the object to be identified and the second interaction information corresponding to the object to be identified; determine the object to be identified according to the object to be identified and the second interaction information Corresponding identification vector set, the identification vector set includes identification vectors corresponding to each second identification information; identify the identification vector set through the information identification model, determine the identity type corresponding to the object to be identified, apply the identification vector obtained by this method The set can fully represent the dynamic changes of the object to be identified in different time periods, and help the model identify the identity type corresponding to the object to be identified.

Description

technical field [0001] The invention relates to the technical field of neural networks, in particular to an object recognition method and device based on a dynamic graph. Background technique [0002] In recent years, neural networks have been widely used in the field of probability prediction. Usually, in order to improve the accuracy of probability prediction, in addition to utilizing the information in the original space domain of features, the interaction information between these features must be efficiently mined. And use the feature interaction information as a basis to identify the feature interaction information through the model to determine the identity type corresponding to the object to be identified. However, when the feature information corresponding to the identity type is relatively complex, it is still difficult to use feature interaction information alone. Improve the accuracy of the identification model for the identification of the object to be identifie...

AI Technical Summary

Problems solved by technology

And use the feature interaction information as a basis to identify the feature interaction information through the model to determine the identity type corresponding to the object to be identified. However, when the feature information corresponding to the identity type is relatively complex, it is still difficult to use feature interaction information alone. Improve the accuracy of the identification model for the identification of the object to be identified

Images