Extract printed English text from an image
Select a screenshot, scan, or photo, choose the layout that most closely matches the page, and start extraction. The result remains editable so you can correct recognition errors before copying or downloading it. The default “one block” mode works well for a paragraph or document crop; automatic mode is better for a page with mixed regions.
Improve OCR accuracy before extraction
Use a sharp, upright image with strong contrast, even lighting, and text large enough to read when zoomed normally. Crop away unrelated graphics when possible. The enlargement option scales smaller images before recognition but does not invent missing detail. Try automatic, column, single-line, or sparse-text layout when the default groups words incorrectly.
Privacy and local storage
The selected image and extracted text are processed in the current browser and are not intentionally uploaded to Gypes. The locally bundled OCR engine can cache its English language model in browser storage so later runs start faster. Clearing site data removes that cache. Ordinary requests for this page and its static files are still handled as described in the privacy policy.
Accuracy limits
This release is optimized for printed English text and common Latin characters. Handwriting, decorative fonts, curved pages, low-resolution photos, complex tables, mathematical notation, vertical text, and other languages can produce omissions or substitutions. Confidence is an engine estimate, not proof that every character is correct.
Always compare important names, account numbers, dates, legal text, medical instructions, and financial values with the original image. Do not discard the source or rely on OCR as the only record.