A new benchmark, ChartDensity-Bench, has been introduced to evaluate the capabilities of multimodal large language models (MLLMs) in reconstructing numerical data from scientific charts, particularly under conditions of high visual density. This benchmark systematically varies the number of charts presented simultaneously to assess model performance across dimensions such as structural reliability, completeness, parseability, and numerical fidelity. Initial experiments reveal that increasing visual density generally degrades the accuracy of numerical reconstruction, with significant variations observed among different MLLMs. AI
IMPACT This benchmark will help researchers develop more robust MLLMs capable of accurately extracting numerical data from complex visual inputs.
RANK_REASON The cluster contains a research paper introducing a new benchmark for evaluating multimodal large language models. [lever_c_demoted from research: ic=1 ai=1.0]
- arXiv
- ChartDensity-Bench
- chart-level reasoning
- MLLMs
- numerical data reconstruction
- Question Answering
- scientific charts
- visual density
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
