Turn a photo or screenshot of a table into an Excel sheet with OCR: rows and columns rebuilt, cells editable before download, XLSX or CSV
Add the photo or screenshot to Image to Excel, choose the language and click Read the table. The OCR engine reads it in your browser and rebuilds the rows and columns; you check and correct any cell on the page, then download .xlsx or .csv. Nothing is uploaded.
Reading the text in a picture of a table with OCR and placing each word in the right row and column of a spreadsheet, so the table can be sorted, summed and edited.
Related terms
Turn a picture of a table into an Excel sheet with OCR, in your browser.
Add the picture
Drop a photo, scan or screenshot of the table (several pictures become several sheets).
Choose the language
Pick the language of the table's text.
Read the table
Click Read the table and wait for the OCR.
Check the cells
Correct any cell on the page, starting with the yellow ones.
Download
Choose .xlsx or .csv and download.
Result: An Excel (.xlsx) or CSV file
The table appears on the page and every cell can be edited. Cells the OCR engine was unsure of are yellow; in the Excel file they are yellow too and carry a note.
Words on the same line form a row and wide gaps split cells. Columns are the stretches most rows have text in, so left-, right- and centre-aligned columns all line up. A title across the table goes into the first column.
The OCR runs in your browser. English and Hindi need nothing from outside AnyTool; other languages download their data file once. The picture never leaves your device.
1,234.56, 1.234,56, 1,23,456.78, (120) and 12% become numbers, and currency symbols are ignored. IDs with a leading zero, like 00123, stay text so the zero is kept.
Choose Chinese (Simplified or Traditional). Its language file downloads once from the jsDelivr CDN; characters are joined without spaces.
Merged cells, text wrapped inside cells and very small print can land in the wrong cell. A straight, sharp screenshot works best; for tables with many empty cells, turn on 'Scattered text'.
Tesseract returns every word with its box. Words whose vertical centres are within about half a text height share a row; a gap wider than about one text height starts a new cell; and columns are found by counting, for each pixel across the table, how many rows have text there.
With 'Numbers as numbers' on, 1,234.56 (US), 1.234,56 (Europe), 1,23,456.78 (India), (120), −120 and 12% become numeric cells; leading-zero IDs stay text. CSV files start with a UTF-8 byte-order mark so Excel opens non-English text correctly.
OCR can misread characters, and merged or multi-line cells may split or join wrongly. Every cell is editable before download, and unsure cells are highlighted.
The OCR runs in the browser; the picture is not uploaded. Only the data file for languages other than English and Hindi is fetched, from the jsDelivr CDN.
| Your table is in | Use | Why |
|---|---|---|
| A photo, scan or screenshot | Image to Excel | OCR reads the text from the picture |
| A PDF with selectable text | PDF to Excel | Reads the text directly, no OCR errors |
| A scanned PDF | PDF to JPG, then Image to Excel | The scan is a picture inside the PDF |
When the PDF has selectable text, PDF to Excel is more accurate than OCR.
