Photograph a price list, invoice or printed table and get editable cells back. Fix anything by clicking it, then download the sheet or paste it straight in.
They are already read in this tab. A pass unlocks every row of every file you convert over the next 7 days, in this browser — and on any other once you sign in.
Numbers, percentages and amounts are written as numbers, so they add up in the spreadsheet.
The image is enlarged if it is small and read by the OCR engine in "sparse" mode, which looks for text anywhere on the page instead of assuming paragraphs. That is what makes ruled tables and shaded headers survive.
Every word comes back with its position. Words are grouped into rows by height, and the columns are found the way you would find them by eye: as the vertical strips of white space that run through most of the rows.
Each cell's text is typed — "1,240" becomes the number 1240, "12.4%" becomes 0.124 formatted as a percentage — and the sheet is written on your machine.
Something else? Write to support@ocrsnip.com.
Really. The engine runs in your browser; the only downloads are the engine and language files from this site, once, cached afterwards. Open the Network tab and convert a table — no request carries the image.
Every table is recognised and shown in full, editable, for free. Downloading the first 10 rows is free. The $4.99 pass unlocks every row of every table — and every bank statement on the statement converter — for 7 days in this browser. One pass, all the tools.
Columns are detected from the gaps between them, so a table whose columns nearly touch may come out merged, and a column with a wide gap inside it may split. A larger or sharper image usually fixes it — and if you are not sure whether the photo is sharp enough to be worth a second try, izntools can score how far out of focus a picture is without uploading it either. Otherwise, the "copy for pasting" button gives you tab-separated text you can split in the spreadsheet itself.
Eleven languages: English, Dutch, French, German, Italian, Polish, Portuguese, Russian, Spanish, Turkish and Ukrainian, each alone or paired with English. The language setting only affects words — headings, names, descriptions. Digits are digits in every one of them, so a table of figures reads the same whichever you pick.
Yes. Thousands separators, decimal commas, currency signs and percent signs are understood, so the cells arrive as numbers you can sum and chart. Anything it cannot read as a number stays as text — it never guesses.
Same engine, its own page: payslip to Excel takes a PDF as well as a photo, and a payslip straight from payroll software is read from the PDF's own text rather than recognised — exact figures, no confidence percentage. It is also the page that says what reading a payslip here does and does not prove.
Then the image to text page is faster: it returns lines of text with no column detection.
ocrsnip