Rapid cross-modal retrieval method and system for incremental data carrying new categories

The invention provides a rapid cross-modal retrieval method and system for incremental data carrying new categories, and the method comprises an incremental hash learning step: extracting a binary representation of a known category label from a known hash code stored in a multimedia known category database, and then according to a similarity relationship between the known category label and an unknown category label, obtaining a binary representation of an unknown category label to supervise the generation of a hash code of incremental data in an incremental category database; and a Hash function learning step: in a Hash function learning process, obtaining an anchor point set from the known category database and the incremental category database through sampling, updating parameters of the deep network by adopting an asymmetric strategy based on the anchor point set, and carrying out Hash function learning to obtain a required model. The method can directly learn the hash code of the unknown incremental data