# RSAgent: Learning to Reason and Act for Text-Guided Segmentation via Multi-Turn Tool Invocations

arXiv: https://arxiv.org/abs/2512.24023 · HTML: https://mlsanity.com/paper/2512.24023/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Placed in Multimodal reasoning, one of 17 directions in LLM reasoning.
- Builds on 18 of its 47 references that are in the map.
- First posted on arXiv 30 Dec 2025.

## Placed in

- LLM reasoning / [Multimodal reasoning](https://mlsanity.com/lr/multimodal-reasoning/)

## Builds on (most cited first, 18 of 18)

- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [Qwen2.5-VL Technical Report](https://mlsanity.com/paper/2502.13923/) (2025)
- [SAM 2: Segment Anything in Images and Videos](https://mlsanity.com/paper/2408.00714/) (2024)
- [Grounded SAM: Assembling Open-World Models for Diverse Visual Tasks](https://mlsanity.com/paper/2401.14159/) (2024)
- [SAM 3: Segment Anything with Concepts](https://mlsanity.com/paper/2511.16719/) (2025)
- [Solving math word problems with process- and outcome-based feedback](https://mlsanity.com/paper/2211.14275/) (2022)
- [Segment Everything Everywhere All at Once](https://mlsanity.com/paper/2304.06718/) (2023)
- [Open-Reasoner-Zero: An Open Source Approach to Scaling Up Reinforcement Learning on the Base Model](https://mlsanity.com/paper/2503.24290/) (2025)
- [DeepEyes: Incentivizing "Thinking with Images" via Reinforcement Learning](https://mlsanity.com/paper/2505.14362/) (2025)
- [Pixel Reasoner: Incentivizing Pixel-Space Reasoning with Curiosity-Driven Reinforcement Learning](https://mlsanity.com/paper/2505.15966/) (2025)
- [Seg-Zero: Reasoning-Chain Guided Segmentation via Cognitive Reinforcement](https://mlsanity.com/paper/2503.06520/) (2025)
- [DeepEyesV2: Toward Agentic Multimodal Model](https://mlsanity.com/paper/2511.05271/) (2025)
- [EVF-SAM: Early Vision-Language Fusion for Text-Prompted Segment Anything Model](https://mlsanity.com/paper/2406.20076/) (2024)
- [VisionReasoner: Unified Reasoning-Integrated Visual Perception via Reinforcement Learning](https://mlsanity.com/paper/2505.12081/) (2025)
- [Seg-R1: Segmentation Can Be Surprisingly Simple with Reinforcement Learning](https://mlsanity.com/paper/2506.22624/) (2025)
- [SAM-R1: Leveraging SAM for Reward Feedback in Multimodal Segmentation via Reinforcement Learning](https://mlsanity.com/paper/2505.22596/) (2025)
- [X-SAM: From Segment Anything to Any Segmentation](https://mlsanity.com/paper/2508.04655/) (2025)
- [Ophiuchus: Incentivizing Tool-augmented "Think with Images" for Joint Medical Segmentation, Understanding and Reasoning](https://mlsanity.com/paper/2512.14157/) (2025)

## Built on by (most cited first, 6 of 6)

- [From Storage to Experience: A Survey on the Evolution of LLM Agent Memory Mechanisms](https://mlsanity.com/paper/2605.06716/) (2026)
- [Vision Harnessing Agent for Open Ad-hoc Segmentation](https://mlsanity.com/paper/2605.19410/) (2026)
- [B-GRTO: Bootstrapped Group Relative Tool Optimization for Referring Segmentation](https://mlsanity.com/paper/2605.23500/) (2026)
- [DGSeg: Dynamic Gating of Semantic-Spatial Guided Predictions for Reasoning Segmentation](https://mlsanity.com/paper/2607.04779/) (2026)
- [InsightSeg: Reusing Correction Insights for Guideline-Consistent Segmentation](https://mlsanity.com/paper/2609.02002/) (2026)
- [Neos: Learning to Interleave Search and Visual Reasoning](https://mlsanity.com/paper/2610.17043/) (2026)

## Abstract (arXiv)

Text-guided object segmentation requires both cross-modal reasoning and pixel grounding abilities. Most recent methods treat text-guided segmentation as one-shot grounding, where the model predicts pixel prompts in a single forward pass to drive an external segmentor, which limits verification, refocusing and refinement when initial localization is wrong. To address this limitation, we propose RSAgent, an agentic Multimodal Large Language Model (MLLM) which interleaves reasoning and action for segmentation via multi-turn tool invocations. RSAgent queries a segmentation toolbox, observes visual feedback, and revises its spatial hypothesis using historical observations to re-localize targets and iteratively refine masks. We further build a data pipeline to synthesize multi-turn reasoning segmentation trajectories, and train RSAgent with a two-stage framework: cold-start supervised fine-tuning followed by agentic reinforcement learning with fine-grained, task-specific rewards. Extensive experiments show that RSAgent achieves a zero-shot performance of 66.5% gIoU on ReasonSeg test, improving over Seg-Zero-7B by 9%, and reaches 81.5% cIoU on RefCOCOg, demonstrating state-of-the-art performance on both in-domain and out-of-domain benchmarks.