VisualToolAgent (VisTA): A Reinforcement Learning Framework for Visual Tool Selection

TL;DR

VisTA introduces a reinforcement learning framework enabling visual agents to dynamically explore and select tools from a diverse library, significantly improving reasoning performance and generalization on visual question-answering benchmarks.

Contribution

It presents a novel RL-based approach for active tool selection in visual reasoning, overcoming limitations of prior prompting and fine-tuning methods.

Findings

Achieves performance gains over baselines on multiple benchmarks.

Enhances generalization to out-of-distribution examples.

Demonstrates effective tool utilization and adaptive reasoning.

Abstract

We introduce VisTA, a new reinforcement learning framework that empowers visual agents to dynamically explore, select, and combine tools from a diverse library based on empirical performance. Existing methods for tool-augmented reasoning either rely on training-free prompting or large-scale fine-tuning; both lack active tool exploration and typically assume limited tool diversity, and fine-tuning methods additionally demand extensive human supervision. In contrast, VisTA leverages end-to-end reinforcement learning to iteratively refine sophisticated, query-specific tool selection strategies, using task outcomes as feedback signals. Through Group Relative Policy Optimization (GRPO), our framework enables an agent to autonomously discover effective tool-selection pathways without requiring explicit reasoning supervision. Experiments on the ChartQA, Geometry3K, and BlindTest benchmarks demonstrate that VisTA achieves substantial performance gains over training-free baselines, especially on out-of-distribution examples. These results highlight VisTA's ability to enhance generalization, adaptively utilize diverse tools, and pave the way for flexible, experience-driven visual reasoning systems.