Mining Fine-Grained Image-Text Alignment for Zero-Shot Captioning via Text-Only Training

TL;DR

This paper introduces a novel zero-shot image captioning method that reduces the modality gap in CLIP's latent space using subregion features, noise injection, and reranking, achieving significant improvements without caption data.

Contribution

The paper proposes a new framework for zero-shot captioning that leverages text-only training and addresses CLIP's modality gap through subregion aggregation and noise strategies.

Findings

Achieves state-of-the-art zero-shot captioning performance on MSCOCO and Flickr30k.

Demonstrates the effectiveness of subregion features and noise injection in reducing modality gap.

Extends the approach to zero-shot VQA, showing its versatility.

Abstract

Image captioning aims at generating descriptive and meaningful textual descriptions of images, enabling a broad range of vision-language applications. Prior works have demonstrated that harnessing the power of Contrastive Image Language Pre-training (CLIP) offers a promising approach to achieving zero-shot captioning, eliminating the need for expensive caption annotations. However, the widely observed modality gap in the latent space of CLIP harms the performance of zero-shot captioning by breaking the alignment between paired image-text features. To address this issue, we conduct an analysis on the CLIP latent space which leads to two findings. Firstly, we observe that the CLIP's visual feature of image subregions can achieve closer proximity to the paired caption due to the inherent information loss in text descriptions. In addition, we show that the modality gap between a paired image-text can be empirically modeled as a zero-mean Gaussian distribution. Motivated by the findings, we propose a novel zero-shot image captioning framework with text-only training to reduce the modality gap. In particular, we introduce a subregion feature aggregation to leverage local region information, which produces a compact visual representation for matching text representation. Moreover, we incorporate a noise injection and CLIP reranking strategy to boost captioning performance. We also extend our framework to build a zero-shot VQA pipeline, demonstrating its generality. Through extensive experiments on common captioning and VQA datasets such as MSCOCO, Flickr30k and VQAV2, we show that our method achieves remarkable performance improvements. Code is available at https://github.com/Artanic30/MacCap.