Research map / Vision-language and multimodal / Multimodal LLMs
ChartCap: Mitigating Hallucination of Dense Chart Captioning
Junyoung Lim, Jaewoo Ahn, Gunhee Kim
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 |
|---|---|---|---|
| fine-tune on ChartCap | Phi3.5-Vision-4B · ChartCap test set | sacreBLEU: 8.41 -> 23.82 | Improved |
| fine-tune on ChartCap | InternVL2.5-8B · ChartCap test set | sacreBLEU: 3.60 -> 19.47 | Improved |
| fine-tune on ChartCap | Phi3.5-Vision-4B · ChartCap test set | OCRScore: 0.4875 -> 0.5179 | Improved |
| fine-tune on 250K original captions instead of ChartCap (ablation) | Phi3.5-Vision-4B · ChartCap test set | sacreBLEU: 8.41 -> 0.09 | Hurt |
| fine-tune on ChartSumm instead of ChartCap (ablation) | Phi3.5-Vision-4B · ChartCap test set | sacreBLEU: 8.41 -> 1.31 | Hurt |
| fine-tune on ChartSumm instead of ChartCap (ablation) | Phi3.5-Vision-4B · VisText test (zero-shot) | Visual Consistency Score: 0.8814 -> 0.8677 | Hurt |
| fine-tune on ChartCap | Qwen2.5-VL-7B · VisText test (zero-shot) | Visual Consistency Score: 0.9044 -> 0.9328 | Improved |
| fine-tune on ChartCap | Qwen2.5-VL-7B · Chart-to-Text PEW subset (zero-shot) | Visual Consistency Score: 0.7739 -> 0.8084 | Improved |
| fine-tune on ChartCap | Phi3.5-Vision-4B · VisText test (zero-shot) | Visual Consistency Score: 0.8814 -> 0.9382 | Improved |
| fine-tune on ChartCap | Phi3.5-Vision-4B · VisText test (zero-shot) | OCRScore: 0.3414 -> 0.3826 | Improved |
| fine-tune on ChartCap | InternVL2.5-8B · VisText test (zero-shot) | OCRScore: 0.3172 -> 0.3360 | Mixed |
| train on ChartCap data rather than ChartSumm captions | captioning dataset · Dataset caption quality (1K samples) | Visual Consistency Score: 0.9008 -> 0.9089 | Improved |
Closest papers by meaning
- ChartNet: A Million-Scale, High-Quality Multimodal Dataset for Robust Chart Understanding (CVPR 2026 · similarity 0.89)
- Visual Self-Refine: A Pixel-Guided Paradigm for Accurate Chart Parsing (ICLR 2026 · similarity 0.87)
- Chart-FR1: Visual Focus-Driven Fine-Grained Reasoning on Dense Charts (CVPR 2026 · similarity 0.87)
- ChartMuseum: Testing Visual Reasoning Capabilities of Large Vision-Language Models (NeurIPS 2025 · similarity 0.87)
- EvoChart: A Benchmark and a Self-Training Approach Towards Real-World Chart Understanding (AAAI 2025 · similarity 0.86)
- Twin-T & TwintVQA: A Reliable Structure-Detail Separating VLM and a Comprehensive Benchmark for Chart and Table Tasks (CVPR 2026 · similarity 0.86)
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.
