Image to CSV (table OCR)
Advertisement
Advertisement
Got a table you can only get as a picture — a screenshot, a photo of a printed spreadsheet, a photographed price list? This tool reads the text with OCR, figures out where the columns are based on how the words line up, and builds a CSV file out of it. A preview shows you the detected rows and columns before you download, so you can check it before trusting it for anything important.
The whole thing runs in your browser: the image is read locally with a WebAssembly OCR engine, and the CSV is generated on your device too — nothing is ever uploaded. Works best with a straight-on, well-lit photo where the columns are reasonably aligned; a tilted or blurry photo will read worse, same as any OCR.
Also useful: Image to text (OCR) · PDF to CSV
Advertisement
Frequently asked questions
What if it doesn’t detect columns correctly?
Table detection is a heuristic based on spacing between words, and it works best on clearly separated columns. If it can’t find a confident column split, it falls back to one column per line of text (with a note telling you that happened) instead of guessing wrong — you always get something usable, worst case a single-column CSV you can split by hand.
Does the CSV open correctly in Excel, with accents intact?
Yes. It’s generated with a UTF-8 byte-order mark, so Excel (including Portuguese and Spanish regional versions), Google Sheets, and LibreOffice all display accented characters correctly — just double-click the file.
Is my image uploaded anywhere?
No. OCR and CSV generation both happen locally via WebAssembly. Images up to 20MB in JPG, PNG, WebP, or BMP; large photos are automatically downscaled before reading. The first use in a browsing session downloads a small OCR engine that then stays cached.