Visual-language pre-training has achieved remarkable success in many multi-modal tasks, largely attributed to the availability of large-scale image-text datasets. In this work, we demonstrate that Multi-modal Large Language Models (MLLMs) can enhance visual-language representation learning by establishing richer image-text associations for image-text datasets. Our approach is simple, utilizing MLLMs to extend multiple diverse captions for each image. To prevent the bias introduced by MLLMs' hallucinations and monotonous language styles, we propose "text shearing" to maintain the quality and availability of extended captions. In image-text retrieval, without introducing additional training cost, our method consistently obtains 5.6 ~ 35.0 and 16.8 ~ 46.1 improvement on Recall@1 under the fine-tuning and zero-shot settings, respectively. Notably, we obtain zero-shot results that are comparable to fine-tuning on target datasets, which encourages more exploration of the versatile use of MLLMs.

翻译：视觉-语言预训练已在多模态任务中取得显著成功，这主要归功于大规模图像-文本数据集的可用性。本研究表明，多模态大语言模型（MLLMs）能够通过为图像-文本数据集建立更丰富的图像-文本关联来增强视觉-语言表示学习。我们的方法简洁高效，利用MLLMs为每张图像扩展生成多样化的描述文本。为防止MLLMs的幻觉现象与单调语言风格带来的偏差，我们提出"文本裁剪"（text shearing）策略以维护扩展描述的质量与可用性。在图像-文本检索任务中，无需引入额外训练成本，我们的方法在微调与零样本设置下，Recall@1指标分别稳定提升5.6~35.0和16.8~46.1。值得关注的是，我们在零样本场景下取得了与目标数据集微调相媲美的结果，这促使我们进一步探索MLLMs的多样化应用潜力。