Unbalanced data-oriented federal cross-modal retrieval method and system

The invention provides a federated cross-modal retrieval method and system oriented to unbalanced data, relates to the field of federated learning and cross-modal retrieval, solves the influence caused by data non-independent identical distribution in a cross-modal retrieval task, and encodes a query sample of a to-be-queried target based on a trained global cross-modal retrieval model so as to obtain a query result of the to-be-queried target. Obtaining a query hash code; performing similarity calculation on the query hash code and a data hash code in the retrieval data set, and obtaining a retrieval result based on the similarity; the global cross-modal retrieval model is obtained based on federal learning training; the method is oriented to non-independent identically distributed data, and feature representation of samples is enriched and enhanced by embedding global feature category prototypes into sample features; semantic information of supervised learning labels is fully utilized, so that the generated