Research map / Vision-language and multimodal / Multimodal LLMs
ChartPoint: Guiding MLLMs with Grounding Reflection for Chart Reasoning
Zhengzhuo Xu, SiNan Du, Yiyan Qi, Siwen Lu, Chengjin Xu, Chun Yuan, Jian Guo
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 PointCoT SFT (ChartPoint) to Qwen2-VL | Qwen2-VL · ChartQA (avg @0.05) | relaxed accuracy: 83.16 -> 85.28 (+2.12%) | Improved |
| add PointCoT SFT (ChartPoint) to Qwen2.5-VL | Qwen2.5-VL · ChartQA (avg @0.05) | relaxed accuracy: 86.36 -> 87.74 (+1.38%) | Improved |
| add PointCoT SFT (ChartPoint) to Qwen2-VL | Qwen2-VL · ChartBench (ALL) | accuracy: 58.90 -> 62.61 (+3.71%) | Improved |
| add PointCoT SFT (ChartPoint) to Qwen2.5-VL | Qwen2.5-VL · ChartBench (ALL) | accuracy: 60.91 -> 65.95 (+5.04%) | Improved |
| add PointCoT SFT to Qwen2.5-VL on extra chart types | Qwen2.5-VL · ChartBench (extra types) | accuracy: 57.26 -> 65.03 (+7.77%) | Improved |
| add Stage1 chart knowledge alignment (ablation) | Qwen2-VL · ChartQA (avg) | relaxed accuracy: +0.58% | Improved |
| add Stage1 chart knowledge alignment (ablation) | Qwen2.5-VL · ChartQA (avg) | relaxed accuracy: +0.14% | No clear effect |
| add text-only CoT data on top of Stage1 (ablation) | Qwen2-VL · ChartBench (ALL) | accuracy: 60.39 -> 60.76 | No clear effect |
| add grounding supervision (PointCoT) instead of text CoT (ablation) | Qwen2-VL · ChartBench (ALL) | accuracy: 60.76 -> 62.61 | Improved |
| add grounding supervision (PointCoT) instead of text CoT (ablation) | Qwen2.5-VL · ChartBench (ALL) | accuracy: 62.98 -> 65.95 | Improved |
| apply PointCoT to Qwen-VL (weak localization) (ablation) | Qwen-VL · ChartBench | accuracy: 28.18 -> 27.92 (-0.26) | Hurt |
| apply PointCoT to ChartMoE (weak localization) (ablation) | ChartMoE · ChartQA | relaxed accuracy: 81.20 -> 81.36 (+0.16) | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- ChartMoE — ChartMoE: Mixture of Diversely Aligned Expert Connector for Chart Understanding (ICLR 2025) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- ChartSketcher: Reasoning with Multimodal Feedback and Reflection for Chart Understanding (NeurIPS 2025 · similarity 0.92)
- ChartNet: A Million-Scale, High-Quality Multimodal Dataset for Robust Chart Understanding (CVPR 2026 · similarity 0.90)
- Chart-FR1: Visual Focus-Driven Fine-Grained Reasoning on Dense Charts (CVPR 2026 · similarity 0.90)
- ChartMuseum: Testing Visual Reasoning Capabilities of Large Vision-Language Models (NeurIPS 2025 · similarity 0.90)
- EvoChart: A Benchmark and a Self-Training Approach Towards Real-World Chart Understanding (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.
