# PixelThink: Towards Efficient Chain-of-Pixel Reasoning

arXiv: https://arxiv.org/abs/2505.23727 · HTML: https://mlsanity.com/paper/2505.23727/

## Key facts

- Cited by 28 papers in the map (data through 14 Oct 2026).
- Placed in Multimodal reasoning, one of 17 directions in LLM reasoning.
- Builds on 72 of its 80 references that are in the map.
- First posted on arXiv 29 May 2025.

## Placed in

- LLM reasoning / [Multimodal reasoning](https://mlsanity.com/lr/multimodal-reasoning/)
- Multimodal LLMs / vision-language models / [VLM benchmarks and visual reasoning](https://mlsanity.com/mllm/vlm-benchmarks-and-visual-reasoning/)

## Builds on (most cited first, 30 of 72)

- [U-Net: Convolutional Networks for Biomedical Image Segmentation](https://mlsanity.com/paper/1505.04597/) (2015)
- [Microsoft COCO: Common Objects in Context](https://mlsanity.com/paper/1405.0312/) (2014)
- [Chain-of-Thought Prompting Elicits Reasoning in Large Language Models](https://mlsanity.com/paper/2201.11903/) (2022)
- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [Segment Anything](https://mlsanity.com/paper/2304.02643/) (2023)
- [Visual Instruction Tuning](https://mlsanity.com/paper/2304.08485/) (2023)
- [Mask R-CNN](https://mlsanity.com/paper/1703.06870/) (2017)
- [DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning](https://mlsanity.com/paper/2501.12948/) (2025)
- [Qwen2.5-VL Technical Report](https://mlsanity.com/paper/2502.13923/) (2025)
- [Qwen2.5 Technical Report](https://mlsanity.com/paper/2412.15115/) (2024)
- [Improved Baselines with Visual Instruction Tuning](https://mlsanity.com/paper/2310.03744/) (2023)
- [Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution](https://mlsanity.com/paper/2409.12191/) (2024)
- [DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs](https://mlsanity.com/paper/1606.00915/) (2016)
- [SAM 2: Segment Anything in Images and Videos](https://mlsanity.com/paper/2408.00714/) (2024)
- [Masked-attention Mask Transformer for Universal Image Segmentation](https://mlsanity.com/paper/2112.01527/) (2021)
- [HybridFlow: A Flexible and Efficient RLHF Framework](https://mlsanity.com/paper/2409.19256/) (2024)
- [s1: Simple test-time scaling](https://mlsanity.com/paper/2501.19393/) (2025)
- [Kimi k1.5: Scaling Reinforcement Learning with LLMs](https://mlsanity.com/paper/2501.12599/) (2025)
- [Modeling Context in Referring Expressions](https://mlsanity.com/paper/1608.00272/) (2016)
- [Grounded SAM: Assembling Open-World Models for Diverse Visual Tasks](https://mlsanity.com/paper/2401.14159/) (2024)
- [Math-Shepherd: Verify and Reinforce LLMs Step-by-step without Human Annotations](https://mlsanity.com/paper/2312.08935/) (2023)
- [LISA: Reasoning Segmentation via Large Language Model](https://mlsanity.com/paper/2308.00692/) (2023)
- [Automatic Chain of Thought Prompting in Large Language Models](https://mlsanity.com/paper/2210.03493/) (2022)
- [An Image is Worth 1/2 Tokens After Layer 2: Plug-and-Play Inference Acceleration for Large Vision-Language Models](https://mlsanity.com/paper/2403.06764/) (2024)
- [Open-Vocabulary Semantic Segmentation with Mask-adapted CLIP](https://mlsanity.com/paper/2210.04150/) (2022)
- [VLM-R1: A Stable and Generalizable R1-style Large Vision-Language Model](https://mlsanity.com/paper/2504.07615/) (2025)
- [Visual-RFT: Visual Reinforcement Fine-Tuning](https://mlsanity.com/paper/2503.01785/) (2025)
- [DriveVLM: The Convergence of Autonomous Driving and Large Vision-Language Models](https://mlsanity.com/paper/2402.12289/) (2024)
- [Stop Overthinking: A Survey on Efficient Reasoning for Large Language Models](https://mlsanity.com/paper/2503.16419/) (2025)
- [LLaVA-PruMerge: Adaptive Token Reduction for Efficient Large Multimodal Models](https://mlsanity.com/paper/2403.15388/) (2024)

## Built on by (most cited first, 28 of 28)

- [Thinking with Images for Multimodal Reasoning: Foundations, Methods, and Future Frontiers](https://mlsanity.com/paper/2506.23918/) (2025)
- [Vision-Language-Action Models for Autonomous Driving: Past, Present, and Future](https://mlsanity.com/paper/2512.16760/) (2025)
- [JarvisEvo: Towards a Self-Evolving Photo Editing Agent with Synergistic Editor-Evaluator Optimization](https://mlsanity.com/paper/2511.23002/) (2025)
- [From Perception to Cognition: A Survey of Vision-Language Interactive Reasoning in Multimodal Large Language Models](https://mlsanity.com/paper/2509.25373/) (2025)
- [Multimodal Referring Segmentation: A Survey](https://mlsanity.com/paper/2508.00265/) (2025)
- [RewardMap: Tackling Sparse Rewards in Fine-grained Visual Reasoning via Multi-Stage Reinforcement Learning](https://mlsanity.com/paper/2510.02240/) (2025)
- [Reinforcing Video Reasoning Segmentation to Think Before It Segments](https://mlsanity.com/paper/2508.11538/) (2025)
- [AVATAR: Reinforcement Learning to See, Hear, and Reason Over Video](https://mlsanity.com/paper/2508.03100/) (2025)
- [Towards Understanding Visual Grounding in Visual Language Models](https://mlsanity.com/paper/2509.10345/) (2025)
- [Look Less, Reason More: Rollout-Guided Adaptive Pixel-Space Reasoning](https://mlsanity.com/paper/2510.01681/) (2025)
- [Efficient Inference for Large Vision-Language Models: Bottlenecks, Techniques, and Prospects](https://mlsanity.com/paper/2604.05546/) (2026)
- [LMMs Meet Object-Centric Vision: Understanding, Segmentation, Editing and Generation](https://mlsanity.com/paper/2604.11789/) (2026)
- [Forging Spatial Intelligence: A Roadmap of Multi-Modal Data Pre-Training for Autonomous Systems](https://mlsanity.com/paper/2512.24385/) (2025)
- [SAM4D: Segment Anything in Camera and LiDAR Streams](https://mlsanity.com/paper/2506.21547/) (2025)
- [DR2Seg: Decomposed Two-Stage Rollouts for Efficient Reasoning Segmentation in Multimodal Large Language Models](https://mlsanity.com/paper/2601.09981/) (2026)
- [Dr. Seg: Revisiting GRPO Training for Visual Large Language Models through Perception-Oriented Design](https://mlsanity.com/paper/2603.00152/) (2026)
- [Discriminative Perception via Anchored Description for Reasoning Segmentation](https://mlsanity.com/paper/2603.04002/) (2026)
- [SAMPO-Path: Segmentation Intent-Aligned Preference Optimization for Pathology Foundation Model Segmentation](https://mlsanity.com/paper/2508.02464/) (2025)
- [CoEditor++: Instruction-based Visual Editing via Cognitive Reasoning](https://mlsanity.com/paper/2603.05518/) (2026)
- [Vision Harnessing Agent for Open Ad-hoc Segmentation](https://mlsanity.com/paper/2605.19410/) (2026)
- [An Efficient Training Pipeline for Reasoning Graphical User Interface Agents](https://mlsanity.com/paper/2511.08172/) (2025)
- [Efficient Reasoning via Thought Compression for Language Segmentation](https://mlsanity.com/paper/2604.02040/) (2026)
- [RCoT-Seg: Reinforced Chain-of-Thought for Video Reasoning and Segmentation](https://mlsanity.com/paper/2605.07334/) (2026)
- [CR-Seg: Attention-Guided and CoT-Enhanced Coarse-to-Refined Reasoning Segmentation](https://mlsanity.com/paper/2606.03564/) (2026)
- [VideoSEG-O3: A Multi-turn Reinforcement Learning Framework for Reasoning Video Object Segmentation](https://mlsanity.com/paper/2606.06819/) (2026)
- [From Structure to Synergy: A Survey of Vision-Language Perception Paradigm Evolution in Multimodal Large Language Models](https://mlsanity.com/paper/2606.26196/) (2026)
- [Seek to Segment: Active Perception for Panoramic Referring Segmentation](https://mlsanity.com/paper/2607.02497/) (2026)
- [DGSeg: Dynamic Gating of Semantic-Spatial Guided Predictions for Reasoning Segmentation](https://mlsanity.com/paper/2607.04779/) (2026)

## Abstract (arXiv)

Existing reasoning segmentation approaches typically fine-tune multimodal large language models (MLLMs) using image-text pairs and corresponding mask labels. However, they exhibit limited generalization to out-of-distribution scenarios without an explicit reasoning process. Although recent efforts leverage reinforcement learning through group-relative policy optimization (GRPO) to enhance reasoning ability, they often suffer from overthinking - producing uniformly verbose reasoning chains irrespective of task complexity. This results in elevated computational costs and limited control over reasoning quality. To address this problem, we propose PixelThink, a simple yet effective scheme that integrates externally estimated task difficulty and internally measured model uncertainty to regulate reasoning generation within a reinforcement learning paradigm. The model learns to compress reasoning length in accordance with scene complexity and predictive confidence. To support comprehensive evaluation, we introduce ReasonSeg-Diff, an extended benchmark with annotated reasoning references and difficulty scores, along with a suite of metrics designed to assess segmentation accuracy, reasoning quality, and efficiency jointly. Experimental results demonstrate that the proposed approach improves both reasoning efficiency and overall segmentation performance. Our work contributes novel perspectives towards efficient and interpretable multimodal understanding. The code and model will be publicly available.