DINO-R1: Incentivizing Reasoning Capability in Vision Foundation Models

TL;DR

DINO-R1 introduces a reinforcement learning-based approach to enhance reasoning capabilities in vision foundation models, achieving superior performance on multiple visual datasets by incentivizing query-based reasoning.

Contribution

It is the first to apply reinforcement learning to incentivize reasoning in vision models, introducing GRQO and stabilizing techniques for improved visual reasoning performance.

Findings

Outperforms supervised baselines on COCO, LVIS, and ODinW datasets.

Achieves strong generalization in open-vocabulary and closed-set scenarios.

Demonstrates the effectiveness of reinforcement learning for reasoning in vision models.

Abstract

The recent explosive interest in the reasoning capabilities of large language models, such as DeepSeek-R1, has demonstrated remarkable success through reinforcement learning-based fine-tuning frameworks, exemplified by methods like Group Relative Policy Optimization (GRPO). However, such reasoning abilities remain underexplored and notably absent in vision foundation models, including representation models like the DINO series. In this work, we propose \textbf{DINO-R1}, the first such attempt to incentivize visual in-context reasoning capabilities of vision foundation models using reinforcement learning. Specifically, DINO-R1 introduces \textbf{Group Relative Query Optimization (GRQO)}, a novel reinforcement-style training strategy explicitly designed for query-based representation models, which computes query-level rewards based on group-normalized alignment quality. We also apply KL-regularization to stabilize the objectness distribution to reduce the training instability. This joint optimization enables dense and expressive supervision across queries while mitigating overfitting and distributional drift. Building upon Grounding-DINO, we train a series of DINO-R1 family models that integrate a visual prompt encoder and a visual-guided query selection mechanism. Extensive experiments on COCO, LVIS, and ODinW demonstrate that DINO-R1 significantly outperforms supervised fine-tuning baselines, achieving strong generalization in both open-vocabulary and closed-set visual prompting scenarios.