Cross-modal retrieval method and device based on discriminative hidden space learning and medium

The invention discloses a cross-modal retrieval method and device based on discriminative hidden space learning and a medium, and the method comprises the steps: extracting a first feature of first modal data and a second feature of second modal data, and constructing a first training set and a second training set; the first training set and the second training set are adopted to train the double-dictionary model added with the discriminative attributes, and a hidden space feature model is obtained; based on a hidden space feature model, respectively projecting the to-be-retrieved modal data features and the modal data features in the retrieval database into a hidden space to obtain corresponding hidden space feature representations; and performing similarity calculation according to the hidden space feature representation to obtain a retrieval result. According to the method and the device, the hidden space is constructed by adopting a double-dictionary learning technology, so that the alignment of the multi