Mllms-augmented Visual-language Representation Learning

Abstract

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 MLL

Stats

• citations0
• S2 citations—
• github stars0
• HF likes0
• heat score0.00
• arxiv keyliu2023mllms

Related papers

• Dreamlip: Language-image Pre-training With Long Captions (2024)10.61
• Breaking The Modality Barrier: Universal Embedding Learning With Multimodal Llms (2025)4.52
• LLM2CLIP: Powerful Language Model Unlocks Richer Cross-modality Representation (2024)2.26
• Indexing Multimodal Language Models For Large-scale Image Retrieval (2026)0.00
• Imagebert: Cross-modal Pre-training With Large-scale Weak-supervised Image-text Data (2020)0.00
• M2-encoder: Advancing Bilingual Image-text Understanding By Large-scale Efficient Pretraining (2024)0.00
• RETLLM: Training And Data-free Mllms For Multimodal Information Retrieval (2026)1.57
• UC2: Universal Cross-lingual Cross-modal Vision-and-language Pre-training (2021)13.05