Multi-class joint subspace learning for cross-modal retrieval

En Yu, Jing Li, Li Wang, Jia Zhang, Wenbo Wan, Jiande Sun

Abstract: Highlights•We propose the multi-class joint subspace learning method for cross-modal retrieval.•The MJSL learns different subspace for different classes and sub-tasks.•The information of semantic overlap among different classes is also explored.•The algorithm outperforms several state-of-the-art cross-modal retrieval algorithms.