space ocr
GuidesArticlesPricingDocs
documents

From Phone Photo to Spreadsheet Row

Snap a photo of any document—receipts, invoices, forms—and get structured data in a spreadsheet. Handles tilted photos and mixed languages automatically.

4 min read· 2026-07-06

You have a stack of receipts or a folder of scanned invoices. The goal is simple: get the key information from each one into a spreadsheet. The reality is tedious. You open an image, squint at the details, and type everything out, field by field. The photos are often taken in a hurry—a little tilted, with weird shadows. It's a slow, error-prone process that no one wants to do.

A document photographed with a phone
A phone photo, tilt and all — still one clean row out.

What if you could skip the manual entry? With space-ocr, you define the columns you need just once—say, Vendor, Invoice Date, and Total Amount. Then, you can upload images of your documents, regardless of their layout. An invoice from one supplier and a receipt from another both resolve to the same, clean row structure in your sheet. It works directly with standard image formats like JPEG, PNG, GIF, and WebP that you get from a scanner or a phone camera.

Upload an image, tell the system what fields you need, and get structured data back.

This works even with the quirks of phone photos. If you take a picture holding your phone vertically, the image file often contains orientation data (EXIF) telling software to rotate it for viewing. space-ocr reads this data on upload, automatically correcting the image so that the coordinates of the extracted text perfectly match the image as you see it on screen. There is no need for manual language selection, either. The system automatically detects and processes text from multiple languages, like Japanese, English, and Korean, all within the same document.

Every extracted value is linked back to its exact location on the original document, providing a clear audit trail.
✓ Verified

To ensure accuracy, space-ocr uses a multi-stage verification process. An LLM first proposes a text value and provides hints about which words it used on the page. The system then performs its own check, matching the proposed value character-by-character against the raw symbols detected by the underlying OCR. This generates a confidence score (match_ratio) and a bounding box (bbox) based on the symbols that actually exist on the page. All coordinates are normalized to a 0–1000 grid, giving you a reliable link between the data and its source.

A handwritten document
Even handwriting and mixed scripts resolve to the same schema.

The pricing is straightforward: $0.05 per image. Every account gets 100 free scans each month to start. And if for some technical reason a scan fails to produce a result, you are not charged for it. It's a simple, pay-as-you-go utility for getting data out of your documents and into a more useful format.

  1. Create a Sheet
    Define the columns you need, like 'Vendor', 'Date', and 'Total'. This is your template.
  2. Take a Photo
    Use your phone to snap a picture of a receipt, invoice, or any document. Don't worry about getting it perfectly flat.
  3. Upload the Image
    Drag and drop the JPEG or PNG file into your Sheet in the space-ocr web app.
  4. Review the Data
    A new row appears with the extracted data, automatically populated in the correct columns. Each value is linked to its location on the photo for verification.
Do I need to straighten or crop my photos?
No. space-ocr automatically handles image rotation based on your phone's camera data (EXIF orientation) and can extract data from tilted or skewed documents.
What file types can I upload?
You can upload common raster image formats like JPEG, PNG, GIF, BMP, TIFF, and WebP. The web app also supports PDFs, which it automatically converts into images page by page for processing.
Does it work with different languages?
Yes. Language detection is fully automatic. The engine is optimized to handle documents with mixed Japanese, Korean, Chinese, and English text, among many others, without any manual settings.
How does it handle different invoice or receipt layouts?
You define your desired schema (your columns) once. The AI then intelligently finds and maps the relevant data from various layouts to fit that single schema.
How much does it cost?
It's $0.05 (or $0.05 / $0.05) per image processed. You also get 100 free scans every month, and we don't charge you if an OCR process fails.
Is the extracted data reliable?
Yes. Every extracted value is verified against the original document's OCR data. It comes with coordinates and a confidence score (match_ratio) so you can always trace the data back to its source on the image.

Stop Typing. Start Uploading.

Get 100 free scans every month. No credit card required to sign up.

Related