BalCapRL: A Balanced Framework for RL-Based MLLM Image Captioning

TL;DR

This paper introduces BalCapRL, a balanced reinforcement learning framework for image captioning with multimodal large language models, optimizing multiple caption quality aspects simultaneously.

Contribution

It proposes a novel multi-objective RL approach with reward normalization and length masking to improve caption quality across several dimensions.

Findings

Consistently improves caption quality metrics across multiple models.

Peak gains of +13.6 in DCScore and +29.0 in CapArena.

Enhanced balance between correctness, coverage, and linguistic quality.

Abstract

Image captioning is one of the most fundamental tasks in computer vision. Owing to its open-ended nature, it has received significant attention in the era of multimodal large language models (MLLMs). In pursuit of ever more detailed and accurate captions, recent work has increasingly turned to reinforcement learning (RL). However, existing captioning-RL methods and evaluation metrics often emphasize a narrow notion of caption quality, inducing trade-offs across core dimensions of captioning. For example, utility-oriented objectives can encourage noisy, hallucinated, or overlong captions that improve downstream question answering while harming fluency, whereas arena-style objectives can favor fluent but generic descriptions with limited usefulness. To address this, we propose a more balanced RL framework that jointly optimizes utility-aware correctness, reference coverage, and linguistic quality. In order to effectively optimize the resulting continuous multi-objective reward formulation, we apply GDPO-style reward-decoupled normalization to continuous-valued captioning rewards and show that it improves performance over vanilla GRPO. Additionally, we introduce length-conditional reward masking, yielding a more suitable length penalty for captioning. Across LLaVA-1.5-7B and Qwen2.5-VL 3B and 7B base models, our method consistently improves caption quality, with peak gains of +13.6 DCScore, +9.0 CaptionQA, and +29.0 CapArena across different models.