Weakly supervised cross-domain alignment with optimal transport
This work addresses a fundamental challenge in vision-language tasks for researchers and practitioners, offering an incremental improvement over existing methods.
The paper tackles cross-domain alignment between images and text under weak supervision, using optimal transport as a regularizer to improve semantic matching, resulting in simpler models achieving comparable or better performance than more complex designs on vision-language tasks.
Cross-domain alignment between image objects and text sequences is key to many visual-language tasks, and it poses a fundamental challenge to both computer vision and natural language processing. This paper investigates a novel approach for the identification and optimization of fine-grained semantic similarities between image and text entities, under a weakly-supervised setup, improving performance over state-of-the-art solutions. Our method builds upon recent advances in optimal transport (OT) to resolve the cross-domain matching problem in a principled manner. Formulated as a drop-in regularizer, the proposed OT solution can be efficiently computed and used in combination with other existing approaches. We present empirical evidence to demonstrate the effectiveness of our approach, showing how it enables simpler model architectures to outperform or be comparable with more sophisticated designs on a range of vision-language tasks.