CURV: Enhancing Chart Understanding Through Curriculum Visual Grounded Reasoning
Quick Answer
CURV introduces a curriculum learning framework for chart question answering (CQA), enhancing intrinsic visual reasoning in multimodal large language models (MLLMs).
Quick Take
It achieves up to 20.50% performance improvement over baselines and generalizes well to real-world benchmarks, validating the effectiveness of dynamic visual grounding.
Key Points
- CURV reformulates CQA as multi-step visual grounded reasoning.
- Introduces CCQA, a three-level curriculum dataset for diverse chart types.
- Demonstrates up to 20.50% improvement over existing models.
- Generalizes to real-world benchmarks with up to 12.30% enhancement.
- Validates dynamic visual grounding for improved chart understanding.
DeepSignal Analysis
What happened
The CURV framework enhances chart question answering (CQA) by integrating visual comprehension with logical reasoning in multimodal large language models (MLLMs). It introduces a three-level curriculum dataset, CCQA, which supports scalable synthetic generation across various chart types. CURV demonstrates significant performance improvements, achieving up to 20.50% better results compared to baseline models.
Key evidence
- CURV reformulates CQA as multi-step visual grounded reasoning, coordinating logical reasoning with dynamic visual grounding.
- The CCQA dataset provides a structured learning path, progressing from basic reasoning tasks to complex multi-chart tasks.
- CURV shows up to 20.50% performance improvement over baselines and generalizes well to real-world benchmarks, achieving up to 12.30% improvement.
Why it matters
The development of CURV addresses critical limitations in current MLLMs, particularly their struggles with visual grounding and coherent reasoning. By enhancing intrinsic visual reasoning capabilities, CURV could lead to more accurate interpretations of charts, which is essential for applications in data analysis and decision-making. The systematic approach to learning through the CCQA dataset may also facilitate broader advancements in multimodal AI.
Paper Resources
📖 Reader Mode
~2 min readAbstract:Chart question answering (CQA) requires multimodal large language models (MLLMs) to integrate visual comprehension with logical reasoning, yet current models struggle with accurate visual grounding and coherent reasoning chains. While extrinsic chain-of-thought prompting and visual cues significantly improve performance, current MLLMs lack intrinsic visual grounded reasoning capabilities, leading to inaccurate perception and reasoning disconnected from visual evidence. To address these limitations, we propose CURV, a curriculum learning framework that develops intrinsic visual reasoning capabilities by reformulating CQA as multi-step visual grounded reasoning, where each step coordinates logical reasoning with dynamic visual grounding through spatial attention concentration. To assist model learning, we further introduce CCQA, a three-level curriculum dataset with scalable synthetic generation across diverse chart types and reasoning patterns. Our curriculum systematically progresses from basic single-operation reasoning to complex multi-chart compositional tasks. Experiments demonstrate that CURV achieves up to $\uparrow20.50\%$ improvements over baselines and is generalizable to real-world benchmarks (up to $\uparrow12.30\%$) and out-of-domain multimodal reasoning tasks (up to $\uparrow10.20\%$), validating the effectiveness of internalizing visual reasoning with dynamic grounding for enhanced chart understanding capabilities. Code is available at: this https URL.
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) |
| Cite as: | arXiv:2608.02833 [cs.CV] |
| (or arXiv:2608.02833v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2608.02833 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Xuehang Guo [view email]
[v1]
Mon, 3 Aug 2026 19:48:00 UTC (37,195 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.CV
See more →ProMoE-FL: Prototype-conditioned Mixture of Experts for Multimodal Federated Learning with Missing Modalities
ProMoE-FL introduces a Prototype-conditioned Mixture-of-Experts framework for multimodal federated learning, effectively addressing missing modalities. It outperforms existing methods on four chest X-ray datasets, demonstrating superior feature synthesis capabilities in both homogeneous and heterogeneous settings.