Modality-Fused Graph Network for Cross-Modal Retrieval

Cross-modal hashing technology has attracted much attention for its favorable retrieval performance and low storage cost. However, existing cross-modal methods, the heterogeneity of data across modalities is still a challenge how to fully explore utilize intra-modality features not been well studied. In this paper, we propose novel approach called Modality-fused Graph Network (MFGN). The network architecture consists text channel an image that are used learn modality-specific features, modality fusion uses graph modality-shared representations reduce modalities. addition, integration module introduced channels features. Experiments on two widely datasets show our achieves better results than state-of-the-art methods.