Drug logistics retrieval method and system based on cross-modality, terminal and storage medium

By employing a cross-modal drug logistics retrieval method, which utilizes image and text feature fusion, graph neural networks, and hash loss optimization, the problems of semantic mismatch and modal imbalance in drug logistics scenarios are solved, achieving high-precision retrieval results.

CN122132505APending Publication Date: 2026-06-02GUANGDONG UNIV OF TECH
0 Cites 1 Cited by

Patent Information

Application Number
CN202610231888.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-02-27
Publication Date
2026-06-02

AI Technical Summary

Technical Problem

Existing technologies for cross-modal data retrieval in pharmaceutical and food logistics scenarios suffer from semantic mismatch, modal imbalance, and difficulty in feature fusion, resulting in low retrieval accuracy.

Method used

A cross-modal drug logistics retrieval method is adopted. By fusing image modal feature representation and text modal feature representation, a label semantic graph is constructed and encoded using a graph neural network. A cross-modal attention mechanism is introduced to construct a hash code and optimize the feature extraction network. The method combines semantic neighborhood perception contrastive hashing and label distribution perception semantic alignment loss to improve retrieval accuracy.

Benefits of technology

It significantly improved the semantic alignment of multimodal features and the generalization ability of the model, thereby enhancing the accuracy of drug logistics retrieval.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention relates to the technical field of logistics management, and discloses a drug logistics retrieval method and system based on cross-modality, a terminal and a storage medium, and the method comprises the steps: taking an encoder as a basis, introducing a feature fusion mechanism, extracting visual representation of text information in a fusion image, and obtaining a fusion image; semantic enhancement is realized through a cross-modal attention mechanism guided by a label atlas; a semantic neighborhood is designed to perceive and compare hash codes, label distribution perceives semantic alignment loss and cosine triple loss, and the discrimination ability of the hash codes is enhanced. According to the method, the semantic alignment degree of the multi-modal features can be improved, the model generalization ability is improved, and the retrieval accuracy of medicine logistics is remarkably improved.
Need to check novelty before this filing date? Find Prior Art