Layout guided masking improves scientific PDF text extraction on CPU
Layout-Guided Masking for GROBID: Lightweight Structural Gains in Large-Scale Scientific PDF Ingestion
Computation and LanguageDigital Libraries
Summary
Turning scientific papers from PDF files into organized, easy-to-read text is hard and slow, especially for large collections. The authors improved a popular tool, GROBID, by adding a simple way to detect parts like figures, tables, and headers on a CPU without needing a graphics processor. This makes the extraction more accurate and faster, keeping costs low while still competing well with GPU-based systems. It works well on biology and materials science papers, making it easier to handle lots of scientific documents.
What this means in practice
- •For digital library teams: Structure large collections of scientific PDFs faster and more accurately without expensive GPUs by integrating layout-guided masking into existing CPU workflows.
- •For data extraction service providers: Deliver cost-efficient, high-quality extraction of tables and figures from scientific PDFs with improved CPU methods to reduce infrastructure costs.
Authors
Luca Foppiano, Sana Khamassi, Vipul Gupta
Abstract
Transforming scholarly PDFs into machine-readable fulltext remains a bottleneck for large-scale information systems. Recent vision-based parsers improve accuracy, but need GPUs and may introduce noise into the extracted text. GROBID, a modular font-stream parser running on CPU, is the de-facto standard for structuring scientific articles and underpins several of the largest open scholarly corpora. We pair it with a lightweight CPU detector localising figure, table, and paratext (header, footer, page number) regions, encoded as typed-area masks whose tokens are routed to GROBID's specialised models or discarded. On two PMC corpora, Bioinformatics (1,926 articles) and Materials Science (2,595), scored against JATS with a section-aware structural protocol, our extension improves over plain GROBID on most metrics (NS $+0.025$/$+0.013$; $+0.086$ paragraph recall on Materials Science, $d_z{=}1.08$), and caption-linked figure recovery improves on both corpora. On the external Table-BRGM benchmark, table detection recovers F1 $0.16 \to 0.94$ and table structure follows (GriTS-Top $0.27 \to 0.78$, below the strongest GPU system). On body text, against four vision-based systems (Docling, MinerU, olmOCR, dots.ocr), it has the best paragraph precision on both corpora, the best section detection on Materials Science, and a character error rate within 0.004 of the best GPU parser. End-to-end on CPU, it costs $2.7$--$3.2\times$ less than the cheapest GPU system (Docling) and $10$--$14\times$ less than generative parsers.