MIRROR: Multimodal Iterative Reasoning via Reflection on Visual Regions

2026-02-21

AI Summary

This work addresses the tendency of existing vision-language models to generate hallucinations when processing ambiguous or complex images, often due to reflection processes that diverge from visual evidence. To mitigate this issue, we propose MIRROR, a novel framework that integrates a region-aware reflection mechanism into a multimodal reasoning loop. MIRROR enables iterative refinement through multiple rounds of draft generation, critique, image-region-based verification, and correction, ensuring that reasoning remains grounded in visual evidence throughout. We also introduce ReflectV, a new dataset designed to support multi-turn reflective training. Experimental results demonstrate that MIRROR significantly improves accuracy on both general and reasoning-oriented vision-language benchmarks while effectively suppressing visual hallucinations.

Technology Category

• Computer Vision: Multi-modal Vision
• Natural Language Processing: Language Grounding & Multi-modal NLP
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

In the era of Vision-Language Models (VLMs), enhancing multimodal reasoning capabilities remains a critical challenge, particularly in handling ambiguous or complex visual inputs, where initial inferences often lead to hallucinations or logic errors. Existing VLMs often produce plausible yet ungrounded answers, and even when prompted to "reflect", their corrections may remain detached from the image evidence. To address this, we propose the MIRROR framework for Multimodal Iterative Reasoning via Reflection On visual Regions. By embedding visual reflection as a core mechanism, MIRROR is formulated as a closed-loop process comprising draft, critique, region-based verification, and revision, which are repeated until the output is visually grounded. To facilitate training of this model, we construct **ReflectV**, a visual reflective dataset for multi-turn supervision that explicitly contains reflection triggers, region-based verification actions, and answer revision grounded in visual evidence. Experiments on both general vision-language benchmarks and representative vision-language reasoning benchmarks show that MIRROR improves correctness and reduces visual hallucinations, demonstrating the value of training reflection as an evidence-seeking, region-aware verification process rather than a purely textual revision step.

Problem

• multimodal reasoning
• visual hallucination
• vision-language models
• region-based verification
• visual grounding

Innovation

• Multimodal Reasoning
• Visual Reflection
• Region-based Verification
• Vision-Language Models
• Iterative Refinement