Improving Cross-Modal Retrieval with Set of Diverse Embeddings
This work addresses the problem of ambiguous cross-modal retrieval for applications like image-text matching, though it appears incremental as it builds on prior set-based embedding approaches.
The paper tackles the challenge of cross-modal retrieval between images and text by proposing a novel set-based embedding method that uses smooth-Chamfer similarity and a set prediction module with slot attention, achieving improved performance on COCO and Flickr30K datasets compared to existing methods.
Cross-modal retrieval across image and text modalities is a challenging task due to its inherent ambiguity: An image often exhibits various situations, and a caption can be coupled with diverse images. Set-based embedding has been studied as a solution to this problem. It seeks to encode a sample into a set of different embedding vectors that capture different semantics of the sample. In this paper, we present a novel set-based embedding method, which is distinct from previous work in two aspects. First, we present a new similarity function called smooth-Chamfer similarity, which is designed to alleviate the side effects of existing similarity functions for set-based embedding. Second, we propose a novel set prediction module to produce a set of embedding vectors that effectively captures diverse semantics of input by the slot attention mechanism. Our method is evaluated on the COCO and Flickr30K datasets across different visual backbones, where it outperforms existing methods including ones that demand substantially larger computation at inference.