Beyond Text and Tables: Vision-Language Model Integration in ComProScanner for Extracting Materials Data from Scientific Figures with High Accuracy
Researchers have developed ComProScanner, an enhanced framework for extracting materials data from scientific literature. This updated version integrates vision-language models (VLMs) to process quantitative data presented in figures, a capability previously lacking in text and table-focused systems. Evaluations using Gemini-3-Flash-Preview demonstrated high accuracy and cost-effectiveness in extracting composition-property pairs from scientific charts and plots. AI
IMPACT Enables more comprehensive automated data extraction from scientific literature, potentially accelerating materials science research.