Research map / Vision-language and multimodal / Multimodal LLMs
StructChart: On the Schema, Metric, and Augmentation for Visual Chart Understanding
Renqiu Xia, Haoyang Peng, Hancheng Ye, Mingsheng Li, Xiangchao Yan, Peng Ye, Botian Shi, Yu Qiao, Junchi Yan, Bo Zhang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add 9K simulated charts (SimChart9K) to ChartQA training (ablation) | StructChart · ChartQA val (strict) | mPrecision: 0.6770 -> 0.7116 | Improved |
| scale simulated data from 0.1K to 9K (ablation) | StructChart · ChartQA val (high tolerance) | mPrecision: 0.8326 -> 0.8527 | Improved |
| add SimChart9K to real merged training data (ablation) | StructChart · PlotQA val (strict) | mPrecision: 0.4549 -> 0.4596 | Improved |
| add SimChart9K to real merged training data (ablation) | StructChart · Chart2Text val (strict) | mPrecision: 0.3156 -> 0.3394 | Improved |
| add SimChart9K to real merged training data (ablation) | StructChart · ChartQA val (high tolerance) | mPrecision: 0.8591 -> 0.8568 | No clear effect |
| merge ChartQA, PlotQA and Chart2Text training sets (ablation) | StructChart · PlotQA val (strict) | mPrecision: 0.1995 -> 0.4549 | Improved |
| use STR instead of LCT as intermediate representation (ablation) | StructChart+Llama3.1-8b · ChartQA test | relaxed accuracy (avg): 55.7 -> 66.4 | Improved |
| use STR instead of LCT as intermediate representation (ablation) | StructChart+Vicuna-13b · ChartQA test | relaxed accuracy (avg): 50.2 -> 59.3 | Improved |
| use STR instead of LCT as intermediate representation (ablation) | StructChart+GPT-4 · ChartQA test | relaxed accuracy (avg): 76.8 -> 78.5 | Improved |
| use STR instead of LCT as intermediate representation (ablation) | StructChart+Llama3.1-70b · ChartQA test | relaxed accuracy (avg): 74.3 -> 75.4 | Improved |
| two-stage StructChart+GPT-4 vs end-to-end GPT-4V | StructChart+GPT-4 · ChartQA test | relaxed accuracy (avg): 75.5 -> 78.5 | Improved |
| use only simulated data without real charts (zero-shot) (ablation) | StructChart · ChartQA val | mPrecision | Hurt |
Closest papers by meaning
- ChartNet: A Million-Scale, High-Quality Multimodal Dataset for Robust Chart Understanding (CVPR 2026 · similarity 0.92)
- EvoChart: A Benchmark and a Self-Training Approach Towards Real-World Chart Understanding (AAAI 2025 · similarity 0.91)
- ChartSketcher: Reasoning with Multimodal Feedback and Reflection for Chart Understanding (NeurIPS 2025 · similarity 0.90)
- ChartMuseum: Testing Visual Reasoning Capabilities of Large Vision-Language Models (NeurIPS 2025 · similarity 0.90)
- VProChart: Answering Chart Question Through Visual Perception Alignment Agent and Programmatic Solution Reasoning (AAAI 2025 · similarity 0.89)
- CharXiv: Charting Gaps in Realistic Chart Understanding in Multimodal LLMs (NeurIPS 2024 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
