Document tool · local browser OCR

Extract text from images and scanned PDFs.

Recognize English and Simplified Chinese in JPG, PNG, WebP, or scanned PDFs, then export editable text. Files are never uploaded to TopicVerge.

Choose or drop an image / PDF

    The self-hosted OCR engine and language model load on demand the first time. Your browser caches model data; files stay on this device.

    Recognized text will appear here and can be edited before download.

    PDF to WordMerge & split PDFMy tool workspace

    What local OCR can and cannot do

    Clear, upright, high-contrast printed text works best. Handwriting, complex tables, curved pages, low-resolution screenshots, and mixed layouts need manual review.

    PDF pages are rendered to images in this browser and recognized one at a time. Processing is limited to the first 10 pages to control memory and wait time. Always compare important OCR output with the source.