Vision-OPD: Learning to See Fine Details for Multimodal LLMs via On-Policy Self-Distillation

TL;DR

Vision-OPD introduces a self-distillation framework that enhances multimodal large language models' ability to focus on fine details in images, improving fine-grained visual understanding without external tools.

Contribution

The paper proposes a novel regional-to-global self-distillation method enabling MLLMs to better perceive relevant evidence in images by internalizing zooming benefits.

Findings

Achieves superior performance on fine-grained visual benchmarks.

Enables models to internalize zooming benefits without external supervision.

Outperforms larger models and existing methods in experiments.

Abstract

Multimodal Large Language Models (MLLMs) still struggle with fine-grained visual understanding, where answers often depend on small but decisive evidence in the full image. We observe a regional-to-global perception gap: the same MLLM answers fine-grained questions more accurately when conditioned on evidence-centered crops than on the corresponding full images, suggesting that many failures stem from difficulty to focus on relevant evidence rather than insufficient local recognition ability. Motivated by this observation, we propose Vision-OPD (Vision On-Policy Distillation), a regional-to-global self-distillation framework that transfers the model's own privileged regional perception to its full-image policy. Vision-OPD instantiates two conditional policies from the same MLLM: a crop-conditioned teacher and a full-image-conditioned student. The student generates on-policy rollouts, and Vision-OPD minimizes token-level divergence between the teacher and student next-token distributions along these rollouts. This enables the model to internalize the benefit of visual zooming without external teacher models, ground-truth labels, reward verifiers, or inference-time tool use. Experiments on multiple fine-grained visual understanding benchmarks show that Vision-OPD models achieve competitive or superior performance against much larger open-source, closed-source, and "Thinking-with-Images" agentic models.