pith. sign in

arxiv: 2505.16192 · v2 · pith:RK65EMHDnew · submitted 2025-05-22 · 💻 cs.CV · cs.AI

VLM-R³: Region Recognition, Reasoning, and Refinement for Enhanced Multimodal Chain-of-Thought

classification 💻 cs.CV cs.AI
keywords textbfreasoningvisualtextualvlm-rchain-of-thoughtemphevidence
0
0 comments X
read the original abstract

Recently, reasoning-based MLLMs have achieved a degree of success in generating long-form textual reasoning chains. However, they still struggle with complex tasks that necessitate dynamic and iterative focusing on and revisiting of visual regions to achieve precise grounding of textual reasoning in visual evidence. We introduce \textbf{VLM-R$^3$} (\textbf{V}isual \textbf{L}anguage \textbf{M}odel with \textbf{R}egion \textbf{R}ecognition and \textbf{R}easoning), a framework that equips an MLLM with the ability to (i) decide \emph{when} additional visual evidence is needed, (ii) determine \emph{where} to ground within the image, and (iii) seamlessly weave the relevant sub-image content back into an interleaved chain-of-thought. The core of our method is \textbf{Region-Conditioned Reinforcement Policy Optimization (R-GRPO)}, a training paradigm that rewards the model for selecting informative regions, formulating appropriate transformations (e.g.\ crop, zoom), and integrating the resulting visual context into subsequent reasoning steps. To bootstrap this policy, we compile a modest but carefully curated Visuo-Lingual Interleaved Rationale (VLIR) corpus that provides step-level supervision on region selection and textual justification. Extensive experiments on MathVista, ScienceQA, and other benchmarks show that VLM-R$^3$ sets a new state of the art in zero-shot and few-shot settings, with the largest gains appearing on questions demanding subtle spatial reasoning or fine-grained visual cue extraction.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 14 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Position Rebinding Cache Reuse: Replay-Free Visual Revisiting for Interleaved Multimodal Reasoning

    cs.CV 2026-06 unverdicted novelty 7.0

    PRCR enables replay-free visual revisiting in interleaved multimodal reasoning by storing raw visual KV caches with spatial coordinates and rebinding keys to position-compatible coordinates, matching replay performanc...

  2. How LLMs See Creativity: Zero-Shot Scoring of Visual Creativity with Interpretable Reasoning

    cs.CL 2026-06 unverdicted novelty 6.0

    Multimodal LLMs match human visual creativity ratings zero-shot on AI images and sketches, with reasoning outputs that are interpretable but do not boost alignment.

  3. Self-Consistent Latent Reasoning: Long Latent Sequence Reasoning for Vision-Language Model

    cs.CV 2026-05 unverdicted novelty 6.0

    SCOLAR fixes information gain collapse in latent visual reasoning by generating independent auxiliary visual tokens via a detransformer, extending acceptable CoT length over 30x and delivering +14.12% gains on reasoni...

  4. Self-Consistent Latent Reasoning: Long Latent Sequence Reasoning for Vision-Language Model

    cs.CV 2026-05 unverdicted novelty 6.0

    SCOLAR addresses information gain collapse in latent visual reasoning by generating independent auxiliary visual tokens from LLM hidden states, extending acceptable CoT length over 30x and achieving +14.12% gains on b...

  5. MHPR: Multidimensional Human Perception and Reasoning Benchmark for Large Vision-Languate Models

    cs.CV 2026-05 unverdicted novelty 6.0

    MHPR is a multidimensional benchmark for LVLM human-centric perception-reasoning with C-RD, SFT-D, RL-D, T-D data tiers and ACVG pipeline, showing training gains on Qwen2.5-VL-7B to near-parity with larger models.

  6. See Further, Think Deeper: Advancing VLM's Reasoning Ability with Low-level Visual Cues and Reflection

    cs.CV 2026-04 unverdicted novelty 6.0

    ForeSight lets VLMs use low-level visual cues and mask-based visual feedback within an RL loop to reason more accurately, with the 7B model beating same-scale peers and some closed-source SOTA on a new benchmark.

  7. CharTool: Tool-Integrated Visual Reasoning for Chart Understanding

    cs.AI 2026-04 unverdicted novelty 6.0

    CharTool equips MLLMs with cropping and code tools plus agentic RL on DuoChart data to raise chart-reasoning accuracy by up to 9.78 percent on benchmarks.

  8. OmniDrive-R1: Reinforcement-driven Interleaved Multi-modal Chain-of-Thought for Trustworthy Vision-Language Autonomous Driving

    cs.CV 2025-12 unverdicted novelty 6.0

    OmniDrive-R1 boosts VLM reasoning score from 51.77% to 80.35% and answer accuracy from 37.81% to 73.62% on DriveLMM-o1 via reinforcement-driven interleaved multi-modal chain-of-thought with annotation-free grounding.

  9. PhyDetEx: Detecting and Explaining the Physical Plausibility of T2V Models

    cs.CV 2025-12 conditional novelty 6.0

    A new dataset and fine-tuned VLM detector/explainer called PhyDetEx shows that current T2V models still struggle to generate videos that obey physical laws, with open-source models performing worse.

  10. DeepEyesV2: Toward Agentic Multimodal Model

    cs.CV 2025-11 unverdicted novelty 6.0

    DeepEyesV2 uses a two-stage cold-start plus reinforcement learning pipeline to produce an agentic multimodal model that adaptively invokes tools and outperforms direct RL on real-world reasoning benchmarks.

  11. How LLMs See Creativity: Zero-Shot Scoring of Visual Creativity with Interpretable Reasoning

    cs.CL 2026-06 unverdicted novelty 5.0

    Multimodal LLMs achieve moderate correlations with human visual creativity ratings in zero-shot evaluation across two datasets, with reasoning outputs providing interpretability but no accuracy gain.

  12. Perceptual Flow Network for Visually Grounded Reasoning

    cs.CV 2026-05 unverdicted novelty 5.0

    PFlowNet decouples perception from reasoning, integrates multi-dimensional rewards with vicinal geometric shaping via variational RL, and reports new SOTA results on V* Bench (90.6%) and MME-RealWorld-lite (67.0%).

  13. When Relations Break: Analyzing Relation Hallucination in Vision-Language Model Under Rotation and Noise

    cs.CV 2026-05 unverdicted novelty 4.0

    Mild rotations and noise significantly increase relation hallucinations in VLMs across models and datasets, with prompt and preprocessing fixes providing only partial relief.

  14. When Relations Break: Analyzing Relation Hallucination in Vision-Language Model Under Rotation and Noise

    cs.CV 2026-05 unverdicted novelty 4.0

    Mild rotations and noise significantly increase relation hallucinations in VLMs across models and datasets, with prompt augmentation and preprocessing offering only partial mitigation.