Invoice OCR that turns supplier bills into data you can trust
Stop retyping invoices. space-ocr reads supplier, numbers, dates, totals, and every line item — and returns each value with its on-page box and a match score.
Every invoice that lands in your inbox is a small data-entry tax. Someone opens the PDF, finds the supplier, the invoice number, the dates, the tax line, the total, then retypes it all into the accounting system — and copies the line items by hand if anyone needs them. It's slow, it's where the typos live, and a single fat-fingered total can hold up a payment run.
Invoice OCR is supposed to take that off your plate: read the bill, get the fields back. The problem with most tools is they hand you a number and ask you to trust it. space-ocr reads the invoice into structured rows and returns every value with the exact spot on the page it came from — a box you can see, plus a score for how well it matched. So before you approve a payment, you can check the figure instead of taking it on faith.
See a real invoice you can check
Hover any field below — the box on the invoice is where that value was read. The supplier, the issue date, the billing period, the due date, the billed amount, the running total, and each line item are all read straight from a real parsed result, not a mockup.
Each value with a box carries a verified on-page location — bbox + 4-point vertices + match_ratio — on a 0–1000 normalized grid (0,0 top-left → 1000,1000 bottom-right), the same shape the live API returns. Hover a field to trace it back to the pixels it came from.
How invoice OCR works in space-ocr
Drop an invoice into the app and it's read into a row: supplier, dates, amounts, and the line items as a sub-table you can sort, filter, and export. A PDF invoice is rendered to an image per page first, then read. If you're calling the API directly, send the page image (the public API takes raster images — JPEG, PNG, GIF, BMP, TIFF, WebP) and you get the same structured result back.
You don't have to describe an invoice from scratch. Pass the built-in templateId invoice, or define your own fields — including an array field whose children describe one line-item row.
curl -s https://api.space-ocr.com/ocr/fields \
-H "Authorization: Bearer $SPACE_OCR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"image": "https://example.com/invoice-page-1.png",
"imageType": "url",
"templateId": "invoice"
}'How to OCR an invoice
- Add the invoiceIn the app, drop the invoice (PDF or image) — each page is rendered to an image and queued for OCR. For AP automation, post it to /upload and get a webhook when it's read.
- Use the invoice templatePass the built-in templateId 'invoice' for supplier, numbers, dates, and totals, or supply your own fields — including an array field with children for the line items.
- Read the structured resultEach value returns with its bbox, vertices, match_ratio, and bbox_source, plus a field_bboxes map locating every field on the invoice.
- Verify before you postClick any amount to highlight the exact region it was read from; a match_ratio below 0.85 flags a value worth a second look. Edits are stored beside the original OCR value.
- Export or queryDownload CSV (UTF-8 BOM, line items unfolded) for your accounting import, or query a stored sheet with GET /view using where, sort, and select — no re-OCR, no extra charge.
Simple, predictable pricing
Pay $0.05 per image (¥10 / ₩100), with a free tier of 100 scans a month and no credit card. Flat plans add monthly scans, more sheets, and storage.
What does invoice OCR pull off an invoice?
Can it read the line items, not just the total?
How do I know the total it read is right?
Can I export invoices to CSV or feed them into accounting?
Does it handle PDF invoices?
How much does invoice OCR cost?
Turn your supplier invoices into checkable data
Free tier — 100 scans a month, no credit card. Every value comes back with its on-page location.