FREE TOOLS
Image Text Search
Recognize image text and find a literal word or phrase, outlining matching recognized words in the image preview.
JPEG, PNG, WebP
TXT, JSON
Browser only
Your input and output are processed in this browser. They are not uploaded for processing. No account is required.
This tool needs JavaScript. If it does not load, refresh the page and try again.
How to use this tool
- Choose a clear, upright image with readable text.
- Choose the document language and automatic or single-block layout.
- Enter a literal word or phrase to find.
- Review the recognized text and word confidence. Download text, the JSON report, or the searchable PDF when available. Use Cancel or Reset to stop and clear the operation.
Examples
An upright white screenshot reading HELLO LOCAL 2026 in large dark letters.
Limitations
JPEG, PNG, or static WebP up to 20 MiB, 12 megapixels and 8,192 pixels per side. Transparent pixels are flattened on white; metadata orientation is removed, so rotate stored pixels first if necessary.
English, French and Dutch models. Printed text works best; handwriting, low resolution, unusual layouts, skew and mixed languages can reduce accuracy. Confidence is an estimate, not a correctness guarantee.
Portable Tesseract 7 LSTM runs in an owned worker. Initial engine and selected models use about 4–11 MiB of downloads; no image/PDF content is sent with these asset requests. Stage timeout: three minutes.
Recognized text is bounded to 5 MiB; searchable PDF to 100 MiB. Text preview shows at most one million characters and confidence table shows the first 200 words; downloads retain complete bounded results.