Image to Excel
Convert image tables to spreadsheets
Extract printed text from a table image, correct the cells, then download Excel or CSV. Check the result against the image before using it.
One PNG, JPG, WebP, or BMP. Up to 20 MB and 20 million pixels. Use a clear, upright image of one table.
Choose a table image to begin.
Review extracted cells
Select a cell to edit it. Row and column actions use the selected cell. Merge joins the selected column with the column to its right across all rows.
Selected: row 1, column A.
Cells export as text to preserve leading zeros. CSV prefixes formula-like values with an apostrophe for spreadsheet safety.
Original OCR text
OCR runs in your browser using English and Indonesian language files. Large images are reduced to 8 million pixels for extraction. The editor supports up to 250 rows, 30 columns, and 3,000 cells. Handwriting, rotated photos, merged cells, and uneven tables may need manual correction.
About Image to Excel
Image to Excel converts printed table text in a picture into an editable grid for CSV or XLSX export. It runs Tesseract OCR in your browser with English, Indonesian, or both languages, then uses the positions of recognized words to infer rows and columns. The inferred structure is a starting point for review. A missing digit or a misplaced column can change the meaning of a price list or report, so compare the cells with the source image before downloading. The editor lets you repair the table without repeating recognition for every correction.
How to Use Image to Excel
Choose a clear table image
Open one JPG, PNG, WebP, or BMP image up to 20 MB, 20 million pixels, and 12,000 pixels on either side. Use a sharp, upright image containing one table. Crop away unrelated headings and nearby text first if they could be mistaken for table content. Images above 8 million pixels are reduced for OCR.
Select the language and extract
Choose English, Bahasa Indonesia, or English + Bahasa Indonesia, then click Extract table. The browser loads the OCR engine and language files and reports recognition progress. You can cancel extraction. If the image is difficult to read or extraction times out, try a smaller, clearer crop rather than relying on a partial result.
Correct the cells and structure
Select a cell to edit its text. Add or remove rows and columns relative to the selected cell. Merge columns joins the selected column with the one to its right across every row. Split column divides values at the first occurrence of your chosen separator, creating one additional column. Check headings, blank cells, decimal marks, and identifiers against the image.
Download the reviewed spreadsheet
Choose Download Excel (.xlsx) or Download CSV after finishing corrections. Excel cells export as text, preserving leading zeros rather than assigning numeric or date types automatically. CSV adds an apostrophe to formula-like values. Keep the image alongside the spreadsheet if you need to check the transcription later.
Common Use Cases
Printed price lists and stock tables
Extract a photographed table of product names, quantities, or reference codes and correct the grid before importing it elsewhere. Look carefully at zeros, the letter O, ones, and the letter I. OCR can confuse similar shapes even when the column layout looks reasonable. For general text without a table, Image to Text provides a more direct text extraction workflow.
Tables captured in screenshots
An upright screenshot with evenly aligned columns gives the inference step clearer positions to work with. Use Crop Image to isolate the table before extraction. If you need calculations or additional spreadsheet work after checking the cells, open the exported file in a spreadsheet editor such as Online Spreadsheet.
Technical Specifications
| Input Formats | JPG, PNG, WebP, BMP |
|---|---|
| Output Formats | XLSX, CSV |
| File Size Limit | 20 MB |
| Processing Engine | 100% Client-side (Runs locally in your browser) |
| Data Retention | Files never leave your device |
| Batch Processing | Single file processing |
Tips for Best Results
Leave a clear gap between columns in the source image when possible. The table inference uses repeated gaps between recognized words, so crowded or irregular spacing can combine unrelated values.
Use the Original OCR text panel to compare the recognized lines with the editable grid. Correcting a structural mistake does not improve the underlying recognition of a blurred character.
The editor allows up to 250 rows, 30 columns, and 3,000 cells, with 2,000 characters per cell. These limits apply together; a table cannot use every row and column maximum at once.
Merged source cells, handwriting, rotated photos, and uneven table layouts may need substantial correction. Work with a smaller region if reconstructing the inferred table would take longer than entering it manually.
Frequently Asked Questions
How does the tool decide where columns begin?
It groups recognized words into lines using their vertical positions, then looks for horizontal gaps repeated across those lines. Those gaps become column boundaries. It does not reconstruct the original spreadsheet or understand the meaning of a heading, and visually aligned results can still contain misplaced values.
Why did my whole table appear in one column?
When the word positions do not support a reliable set of boundaries, the tool falls back to one recognized line per row in a single column. Use Split column, Add column, and cell editing to rebuild it, or retry with a clearer crop and more consistent spacing.
Does Split column separate every space in a cell?
No. It splits at the first selected separator in each cell of the selected column. The choices are space, comma, semicolon, vertical bar, and tab. The rest of the value goes into the new column. Review all rows because the same separator can occur in different places.
Will Excel recognize the extracted prices as numbers?
XLSX exports every edited value as text. This preserves values such as product codes with leading zeros, but you will need to convert appropriate cells to numbers or dates in your spreadsheet application before calculations. OCR does not supply formulas, formatting, or the original workbook structure.
Are image contents sent to an OCR server?
Recognition runs in your browser. The page loads its Tesseract engine and English or Indonesian language assets from the site, then processes the selected image locally. Browser memory and device speed affect extraction; a large or difficult image can fail or take longer to finish.
Can I extract several images or a PDF at once?
This tool accepts one supported image at a time. It does not accept a PDF or create a workbook containing several extracted images. Download the corrected table before clearing it or choosing another source, and combine the results later in a spreadsheet application if needed.