XLsheetAI
How to Convert an Image to Excel: 5 Methods That Actually Work
- The fastest way to convert an image to Excel on a phone is an AI table-extraction app: photograph the table, let the AI rebuild the grid, review, export as XLSX or CSV.
- If you already pay for Microsoft 365, Excel's built-in "Data from Picture" does the same job inside the Excel mobile app and Excel for Windows.
- Free fallbacks exist: Google Drive's OCR gets the text out but usually loses the table structure, and Excel's Get Data > From PDF handles PDF tables well.
- No method is perfect. Clean printed tables extract reliably; handwriting, blur, and merged cells need a review pass before you trust the numbers.
You have a table trapped in a photo, a screenshot, or a scanned page, and you need it in a spreadsheet. Retyping 20 rows of numbers is slow and error-prone. The good news: turning a picture into an editable Excel sheet is now a solved problem for most real-world tables. This guide covers every legitimate method, compares them honestly, and shows how to fix the extraction failures that trip people up.
What "convert image to Excel" actually involves
Converting an image to Excel means running the picture through optical character recognition (OCR) plus table detection, so software can read the characters and rebuild the row-and-column structure as real spreadsheet cells. Simple OCR only gives you a blob of text. Table-aware extraction, which every method below attempts, gives you an editable grid.
That distinction matters. A tool that reads the text but drops the structure leaves you re-splitting lines into columns by hand, which can take almost as long as typing. Judge any picture to Excel tool on two things: character accuracy, and how faithfully it preserves rows, columns, and headers.
Method 1: A phone app with AI table extraction (XLsheetAI)
When the table exists on paper, the most convenient method is a phone app built for the job. XLsheetAI's Tools tab converts a photo or screenshot of a table into an editable grid you can fix up on the spot, then export as XLSX, CSV, or PDF. It suits receipts, price lists, printed reports, and reasonably neat handwritten tables.
- Install XLsheetAI and open the Tools tab. The image to Excel converter lives there, alongside the other extraction tools.
- Capture or choose the image. Photograph the table directly, or pick an existing photo or screenshot from your gallery.
- Crop to the table. Trim away surrounding paragraphs, margins, and background clutter. The tighter the crop, the cleaner the detection.
- Run the AI extraction. The app detects rows, columns, and cell values and rebuilds them as a grid. This step needs an internet connection, since the AI runs on a server, not on your phone.
- Review and edit the grid. Scan the headers and numbers in the editable grid and correct any misreads before you rely on them.
- Export. Save as XLSX to open in Excel or Google Sheets, CSV for imports, or PDF for sharing a read-only copy.
Extraction quality is decided mostly at step 2, before any AI is involved. A few seconds of care with the photo saves minutes of cell-by-cell correction later:
- Lay the page on a flat surface so the rows stay straight.
- Use even light. Daylight from a window beats an overhead bulb that casts your phone's shadow across the page.
- Fill the frame with the table. Small, distant tables mean small, ambiguous characters.
- Shoot straight-on, not at an angle. Perspective skew bends columns and confuses the grid detection.
- Keep your own shadow off the paper.
- Crop to the table before converting, so headers and footers on the page do not get pulled in as extra rows.
Have a table stuck in a photo right now? XLsheetAI's Tools tab turns pictures of tables into editable spreadsheets you can export as XLSX, CSV, or PDF.
Method 2: Excel's built-in "Data from Picture"
Microsoft ships its own picture to Excel feature. "Data from Picture" is available in the Excel mobile apps for Android and iOS, and in Excel for Windows through a Microsoft 365 subscription. It photographs or imports an image, extracts the table, lets you review flagged cells, then inserts the result directly into your worksheet.
On mobile, tap the sheet, choose Data from Picture, and take or select a photo. On Windows with Microsoft 365, it lives under Data > From Picture, pulling from a file or a screen clipping. Either way, Excel highlights cells it is unsure about for you to confirm before inserting, a genuinely useful review step.
The caveats: the Windows version requires a Microsoft 365 subscription, since older perpetual licenses such as Excel 2019 and 2021 do not include it, and the review flow gets fiddly on a small phone screen for wide tables. Accuracy on clean printed tables is high. If your workflow already ends in desktop Excel, this is the natural choice; if the data starts on paper and needs to travel as a file, a phone app that exports directly is often quicker.
Method 3: Google Drive OCR into Docs, then paste into Sheets
Google Drive offers free OCR: upload the image, right-click it, and choose Open with > Google Docs. Drive runs recognition and produces a Doc containing the image plus the extracted text underneath. You then copy that text into Google Sheets, splitting it into columns if needed.
The price is right and the character recognition is decent, but there is a structural catch: Drive's OCR is built for documents, not tables, so output often arrives as plain lines with the column boundaries gone. A simple two-column list can be rebuilt in Sheets with Data > Split text to columns. Anything with five columns, blank cells, or comma-bearing numbers gets painful fast.
Use this route for small, simple tables when you would rather tidy for a few minutes than install anything. Skip it for financial tables, where a silently dropped column boundary can shift numbers into the wrong field.
Method 4: Desktop PDF and OCR tools
If your table lives in a PDF rather than a photo, you may not need image conversion at all. Excel for Windows can import PDF tables directly through Data > Get Data > From File > From PDF, which lists every table Power Query finds in the document and loads your selection straight into the grid.
This works brilliantly for "native" PDFs, the kind exported from software, because the text is already digital and there is no recognition step to get wrong. Bank statements and report exports usually qualify, and it runs locally with no internet needed. Scanned PDFs are different: they are just pictures wrapped in a PDF, so From PDF finds no text and you are back to OCR. Desktop tools such as Adobe Acrobat's export-to-Excel handle those, generally at a subscription price.
Rule of thumb: try selecting text in the PDF. If you can highlight it, use Get Data > From PDF. If you cannot, treat it as an image and use methods 1 to 3.
Method 5: Manual retyping (sometimes it really is faster)
For tiny tables, typing wins. If you are facing five rows and three columns, you will finish keying them in before any app has opened its camera, and you get a built-in accuracy check because your eyes touch every value. No upload, no review pass, no cleanup.
The crossover sits around 10 to 15 rows for most people. Below that, type. Above it, extraction pays for its overhead quickly: a 50-row price list is a five-minute extract-and-review job versus a half hour of error-prone typing. Manual entry is also the only fully offline option, and the only one where the data never leaves your hands.
Which method should you use? Honest comparison
No single method wins everywhere. Phone AI apps and Excel's Data from Picture lead for photographed paper tables, Get Data > From PDF wins for digital PDFs, Drive OCR wins on price, and typing wins for tiny tables. Match the tool to where your table lives and how many rows you face.
| Method | Best for | Accuracy on clean tables | Needs internet | Cost |
|---|---|---|---|---|
| XLsheetAI phone app | Photos of paper tables on the go; direct XLSX/CSV/PDF export | High; review handwriting and blur | Yes | Free tier; subscription for heavy use |
| Excel Data from Picture | People already working in Excel mobile or Microsoft 365 on Windows | High, with a built-in cell review step | Yes | Included with Microsoft 365 |
| Google Drive OCR | Small, simple tables on a zero budget | Medium; text is read well but structure is often lost | Yes | Free |
| Excel Get Data > From PDF | Native (text-based) PDF reports and statements | Very high; no recognition step for native PDFs | No | Included with Excel for Windows |
| Manual retyping | Tiny tables, roughly under 10-15 rows; confidential data | As accurate as you are careful | No | Free |
Why extraction fails, and how to fix it
Almost every bad conversion traces back to a photo problem, an ambiguous character, or a table structure the AI cannot flatten cleanly. Clean printed tables extract reliably; handwriting, blur, and merged cells are where errors concentrate. The table below covers the most common failures and the practical fix for each.
| Problem | Cause | Fix |
|---|---|---|
| Merged cells split or duplicated | A merged header spans several columns, which has no clean single-cell equivalent | Expect merged headers to land in one cell or repeat; unmerge and retitle columns after export |
| Random wrong characters | Low light or motion blur makes 8 look like B, 5 like S | Reshoot on a flat surface in daylight, rest your elbows to steady the phone |
| Rows bleeding into each other | Photo taken at an angle, so gridlines converge | Shoot straight-on from directly above; small skew is corrected automatically, steep angles are not |
| Handwritten digits misread | Recognition models are trained mostly on print; a handwritten 1, 7, or 4 can look alike | Budget a full review pass; check every total against the original |
| Currency symbols mangled or attached to numbers | ₹, €, and $ sit flush against digits and get read as part of the value | Strip symbols after export and apply a currency number format in Excel instead |
| Decimals and thousands separators swapped | Indian format (1,00,000) and European format (1.000,50) differ from US comma conventions | Spot-check magnitudes after import; set your locale in Excel, or find-and-replace separators before formatting as numbers |
| Multi-page tables arrive fragmented | Each photo is processed as its own table with its own header row | Convert one page at a time, delete repeated headers, then stack the pages with copy-paste or Power Query append |
Post-conversion checklist: trust, but verify
Never ship an extracted table without a verification pass. Two minutes of smart sampling catches the silent errors, the ones that look plausible, before they reach a report or an invoice.
- Verify totals. If the original shows column totals, add a quick
SUMand compare. A mismatch means at least one misread cell. - Check number formatting. Confirm values are stored as numbers, not text, and that decimal separators match your locale.
- Spot-check three random rows against the original image, every column, not just the first.
- Watch for O vs 0 and 1 vs l in codes, IDs, and part numbers, where a wrong character will not show up in any total.
Once the data is clean, the usual next step is analysis. If you are matching the imported rows against another sheet, our XLOOKUP vs VLOOKUP guide covers which lookup to reach for, and the Excel formulas cheat sheet has the SUMIFS and COUNTIFS patterns you will want for summarizing it.
Frequently asked questions
Can I convert a photo of a handwritten table to Excel?
Yes, but expect to proofread. AI extraction handles neat, printed-style handwriting reasonably well, while cursive or cramped digits often get misread. Take the sharpest photo you can, then check every number before using the sheet. For messy handwriting, extracting the structure and correcting values by hand still beats retyping everything.
Does image to Excel conversion work offline?
No. AI-based extraction, including XLsheetAI, Excel's Data from Picture, and Google Drive OCR, processes images on remote servers, so you need an internet connection. Manual retyping is the only fully offline method. If you shoot photos somewhere without signal, save them and run the conversion once you are back online.
Can I export the extracted table to CSV or PDF?
Yes. XLsheetAI lets you edit the extracted grid in the app and export it as XLSX, CSV, or PDF, so you can open the result in Excel or Google Sheets, or share a read-only copy. Excel's Data from Picture inserts data directly into a workbook, which you can then save in any format.
Is my data private, or is the image uploaded somewhere?
The image is uploaded for processing, because AI table extraction runs on servers rather than on your phone. That applies to every method covered here except manual typing. For sensitive documents such as payroll or medical records, check your organisation's data policy first, or retype the confidential columns yourself.
Do screenshots convert better than photos?
Usually, yes. Screenshots are perfectly flat, evenly lit, and free of perspective distortion, so a clean screenshot of a table extracts very reliably. Photos add angle, shadow, and blur. If the table is already on a screen, take a screenshot instead of photographing the monitor, and crop it tightly to the table.
Stop retyping tables. Snap a photo, get an editable spreadsheet, export to XLSX, CSV, or PDF. XLsheetAI also generates and explains formulas when you are ready to work with the data.