Image to Excel

How to convert Image to Excel

1. Add an image of a table. A scan, a photo, or a screenshot.
2. Check the detected grid. Rows and columns are inferred from where the text sits.
3. Download the .xlsx. A real workbook that opens in Excel, Sheets, and Numbers.

About Image to Excel

This is an image to Excel converter: it takes a picture of a table and gives back a spreadsheet you can edit. Convert an image to Excel when the data exists only as a photograph: a receipt, a printed report, a screenshot of a table you cannot select.

This converts a photo, picture or scanned table into a real spreadsheet: an .xlsx file that Excel, Google Sheets, LibreOffice and Numbers all open directly. Not a CSV with the extension changed, and not the old .xls format, which nothing has produced by default in fifteen years. If you want plain comma-separated data for a script instead, Image to CSV is the same recognition with a different output.

This is two hard problems stacked on each other. First the text has to be recognised from an image, which is OCR. Then the recognised words have to be arranged back into rows and columns based on where they sat on the page, which is table reconstruction. Each step introduces its own errors, and they compound, so treat the output as a strong starting point rather than a finished spreadsheet.

The tool runs Tesseract over the image and keeps the bounding box of every recognised word. Words are grouped into rows by vertical position, and column boundaries are inferred from the x positions where words consistently start down the page. Those become spreadsheet columns. The result is a genuine .xlsx workbook, an Office Open XML package zipped in the browser, not a CSV with a renamed extension.

Input quality decides everything here, more than for plain OCR. A table photographed at an angle has converging column edges, which destroys the position-based column detection even when every individual word is recognised correctly. Shoot square on. Crop to just the table before running. Make sure the whole grid is in frame and reasonably lit. Two minutes spent on the photograph is worth more than any amount of fiddling afterwards.

Always check the preview against the source before relying on the numbers. Merged cells, wrapped text spilling onto a second line, and columns where figures are right-aligned under a left-aligned header are the usual failure modes. If your table is in a PDF that has a text layer rather than being a scan, use <a href="/pdf-to-excel">PDF to Excel</a> instead: skipping the recognition step removes an entire category of error.

Want plain data instead of a workbook? PNG to CSV handles screenshots and JPG to CSV handles photos. For code, PNG to JSON keeps every word with its position. For the whole process, from capture to checking the result, see how to get a table out of a screenshot.

Photo to spreadsheet, without retyping

Most people land here holding a phone photo of a printed table: a price list, a timetable, a page of results. The aim is a spreadsheet you can sort and sum. That is what this produces, and it is also how you extract a table from an image that came from somewhere else, such as a chat or a slide. If the table is on your own screen, take a screenshot instead of a photo. It will come out noticeably cleaner. The full process is in how to get a table out of a screenshot.

Image to Excel FAQ