SwAMP: Swapped Assignment of Multi-Modal Pairs for Cross-Modal Retrieval

TL;DR

This paper introduces SwAMP, a novel cross-modal retrieval method that uses self-labeled swapped class assignments to improve semantic alignment across modalities, outperforming traditional contrastive learning.

Contribution

The paper proposes a new loss function based on self-labeling and swapping class labels to enhance cross-modal retrieval performance.

Findings

Significant improvement over contrastive learning in multiple retrieval tasks

Effective semantic alignment across modalities using swapped pseudo labels

Applicable to text-video, sketch-image, and image-text retrieval

Abstract

We tackle the cross-modal retrieval problem, where learning is only supervised by relevant multi-modal pairs in the data. Although the contrastive learning is the most popular approach for this task, it makes potentially wrong assumption that the instances in different pairs are automatically irrelevant. To address the issue, we propose a novel loss function that is based on self-labeling of the unknown semantic classes. Specifically, we aim to predict class labels of the data instances in each modality, and assign those labels to the corresponding instances in the other modality (i.e., swapping the pseudo labels). With these swapped labels, we learn the data embedding for each modality using the supervised cross-entropy loss. This way, cross-modal instances from different pairs that are semantically related can be aligned to each other by the class predictor. We tested our approach on several real-world cross-modal retrieval problems, including text-based video retrieval, sketch-based image retrieval, and image-text retrieval. For all these tasks our method achieves significant performance improvement over the contrastive learning.