Cross-Modal Denoising: A Novel Training Paradigm for Enhancing Speech-Image Retrieval

TL;DR

This paper introduces a cross-modal denoising training paradigm that improves speech-image retrieval by enabling finer alignment between modalities, outperforming existing methods without increasing inference complexity.

Contribution

We propose a novel cross-modal denoising task during training that enhances fine-grained alignment in speech-image retrieval, without affecting inference speed.

Findings

Outperforms state-of-the-art by 2.0% in mean R@1 on Flickr8k

Achieves 1.7% improvement in mean R@1 on SpokenCOCO

Effective in capturing fine-grained cross-modal details

Abstract

The success of speech-image retrieval relies on establishing an effective alignment between speech and image. Existing methods often model cross-modal interaction through simple cosine similarity of the global feature of each modality, which fall short in capturing fine-grained details within modalities. To address this issue, we introduce an effective framework and a novel learning task named cross-modal denoising (CMD) to enhance cross-modal interaction to achieve finer-level cross-modal alignment. Specifically, CMD is a denoising task designed to reconstruct semantic features from noisy features within one modality by interacting features from another modality. Notably, CMD operates exclusively during model training and can be removed during inference without adding extra inference time. The experimental results demonstrate that our framework outperforms the state-of-the-art method by 2.0% in mean R@1 on the Flickr8k dataset and by 1.7% in mean R@1 on the SpokenCOCO dataset for the speech-image retrieval tasks, respectively. These experimental results validate the efficiency and effectiveness of our framework.