Layout analysis
Layout analysis is the part of document parsing that finds the regions on a page, such as titles, paragraphs, tables, figures, captions and footers, labels each one, and works out the order a person would read them in.
Also called: reading order, layout detection
By Divya Sharma · Checked Nov 15, 2026 · How we review
Text pulled straight from a PDF often comes out in the wrong order. Two columns get interleaved line by line, footnotes land mid-sentence, and page headers repeat on every page. Layout analysis fixes this before the text reaches chunking or a model, which is why it matters for any document parsing pipeline.
How it works
Most systems treat it as object detection on the page image. A model draws bounding boxes around regions and assigns each a class. The DocLayNet dataset, for example, labels every region with one of 11 classes such as caption, footnote, formula, list item, table, title and text. A second step sorts the boxes into reading order, handling columns, sidebars and figures. The detected regions then go to the right tool: text blocks to OCR if needed, tables to table extraction, figures to an image model.
Example
Given a two-column research paper, layout analysis finds the title, abstract, the two text columns, a figure with its caption and a references list. Reading order puts the left column before the right, keeps the caption with its figure, and drops the running header. Open-source parsers such as Docling include this step.
Limits
Unusual designs, such as magazines, forms, posters or slides, fall outside what most models were trained on, and reading order across complex pages is still error-prone. A vision-language model can handle layout implicitly but gives less control over the result.
Related terms
- Document parsing
- Document parsing converts files such as PDFs, scans, slides and spreadsheets into structured text that software can use, keeping headings, reading order, tables and images instead of a flat stream of characters.
- Optical character recognition
- Optical character recognition (OCR) is software that reads text from images, such as scanned pages, photos or screenshots, and turns it into machine-readable characters that can be searched, copied or fed to a model.
- Table extraction
- Table extraction is the task of finding tables in documents and rebuilding their rows, columns, headers and cells as structured data, such as HTML, Markdown or CSV, instead of loose lines of text.
- Vision-language model
- A vision-language model (VLM) is a model that takes images and text as input and produces text, so it can describe pictures, answer questions about them, read documents and extract data from screenshots.
- Chunking
- Chunking is the step that splits long documents into smaller passages before they are embedded and indexed, so a retrieval system can find and return only the parts of a document that answer a query.