Pith. sign in

REVIEW 6 cited by

Unsupervised Visual Chain-of-Thought Reasoning via Preference Optimization

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2504.18397 v2 pith:A5JWW6FM submitted 2025-04-25 cs.CV

Unsupervised Visual Chain-of-Thought Reasoning via Preference Optimization

classification cs.CV
keywords uv-cotvisualreasoningpreferencedatamllmboundingbounding-box
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Chain-of-thought (CoT) reasoning greatly improves the interpretability and problem-solving abilities of multimodal large language models (MLLMs). However, existing approaches are focused on text CoT, limiting their ability to leverage visual cues. Visual CoT remains underexplored, and the only work is based on supervised fine-tuning (SFT) that relies on extensive labeled bounding-box data and is hard to generalize to unseen cases. In this paper, we introduce Unsupervised Visual CoT (UV-CoT), a novel framework for image-level CoT reasoning via preference optimization. UV-CoT performs preference comparisons between model-generated bounding boxes (one is preferred and the other is dis-preferred), eliminating the need for bounding-box annotations. We get such preference data by introducing an automatic data generation pipeline. Given an image, our target MLLM (e.g., LLaVA-1.5-7B) generates seed bounding boxes using a template prompt and then answers the question using each bounded region as input. An evaluator MLLM (e.g., OmniLLM-12B) ranks the responses, and these rankings serve as supervision to train the target MLLM with UV-CoT by minimizing negative log-likelihood losses. By emulating human perception--identifying key regions and reasoning based on them--UV-CoT can improve visual comprehension, particularly in spatial reasoning tasks where textual descriptions alone fall short. Our experiments on six datasets demonstrate the superiority of UV-CoT, compared to the state-of-the-art textual and visual CoT methods. Our zero-shot testing on four unseen datasets shows the strong generalization of UV-CoT. The code is available in https://github.com/kesenzhao/UV-CoT.

discussion (0)

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

Forward citations

Cited by 6 Pith papers

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

  1. HiDe: Rethinking The Zoom-IN method in High Resolution MLLMs via Hierarchical Decoupling

    cs.CV 2025-09 unverdicted novelty 7.0

    HiDe is a training-free hierarchical decoupling method that separates key visual tokens from background interference in high-resolution MLLMs to achieve new state-of-the-art results on V*Bench, HRBench4K, and HRBench8K.

  2. GRIT: Teaching MLLMs to Think with Images

    cs.CV 2025-05 unverdicted novelty 7.0

    GRIT introduces a grounded reasoning paradigm for MLLMs where reasoning chains interleave text and bounding boxes, trained via GRPO-GR reinforcement learning on as few as 20 examples without annotations.

  3. HART: High-Resolution Annotation-Free Reasoning Technique through a Closed-loop Framework

    cs.CV 2026-02 conditional novelty 6.0

    HART uses a closed-loop 'crop-and-answer' training scheme plus a dynamic-weight GRPO variant to improve LMM grounding and high-resolution reasoning without bounding-box annotations.

  4. CropVLM: Learning to Zoom for Fine-Grained Vision-Language Perception

    cs.CV 2025-11 unverdicted novelty 6.0

    CropVLM uses reinforcement learning to learn image zooming policies that boost fine-grained perception in VLMs on out-of-domain high-resolution tasks without labeled boxes, synthetic data, or VLM changes.

  5. HiDe: Rethinking The Zoom-IN method in High Resolution MLLMs via Hierarchical Decoupling

    cs.CV 2025-09 conditional novelty 6.0

    HiDe uses token-wise attention decoupling and layout-preserving decoupling to build compact crops that push Qwen2.5-VL and InternVL3 to state-of-the-art scores on high-resolution VQA benchmarks.

  6. LaV-CoT: Language-Aware Visual CoT with Multi-Aspect Reward Optimization for Real-World Multilingual VQA

    cs.CV 2025-09 unverdicted novelty 6.0

    LaV-CoT introduces a multi-stage visual CoT pipeline and GRPO training with language-consistency rewards, delivering up to 9.5% accuracy gains on multilingual VQA benchmarks over similar-sized open models.