Guides & articles

Why document AI needs an audit trail, and how per-field bounding boxes make every extracted value checkable against the source image.

A practical tutorial: capture a document image, define your columns once, upload to auto-fill rows, and export a clean UTF-8 CSV — line items and all.

A developer guide to the space-ocr API for extracting data from invoices — curl and Python examples, invoice templates, custom field schemas, and bounding-box provenance.

What to look for in receipt and invoice OCR — verifiable accuracy, line items, export, API, and an audit trail — and how space-ocr delivers each, proven with a live demo.

Turn the hell of typing receipts by hand into nothing more than snapping a photo and exporting a CSV. Line items expand to one row per product, UTF-8 BOM means no garbled text, and every value can be checked against the original, so the data drops straight into freee / Money Forward / Yayoi.

Turn invoices and delivery notes into structured data with a single API call. A developer-focused walkthrough of POST /ocr/fields — curl/Python examples, templates and custom fields, exploding line items into rows, per-value bbox verification, and CSV export.

Use per-field bounding boxes and match ratios to spot-check extracted data fast — without re-running OCR.

Declare the line-item table as an array field and get one structured, verifiable row per item — then expand it straight into CSV.

A scanned page is an image, not a spreadsheet — here's how to read it into structured rows and export a CSV that opens directly in Excel.

When a Textract alternative makes sense — verifiable per-value coordinates, Japanese/Korean/Chinese support, a queryable sheet, and flat per-image pricing with no AWS account — and where Textract still shines.

Most OCR APIs return bounding boxes — but coordinate systems differ and a box only says where, not how sure. How to get source coordinates plus a per-value match ratio you can verify.

Google Cloud Vision is excellent raw OCR — text plus pixel boundingPoly and a recognition confidence — but structured key-value fields are a separate product (Document AI). Here's the honest dividing line, with a live, checkable demo.

Receipt OCR three ways: self-hosted Tesseract, cloud Google Vision, and verification-first space-ocr. Where each wins — pixel boxes vs. structured fields, recognition confidence vs. character-coverage match ratio — proven with a live demo.

An OCR API where every value carries source coordinates and a match_ratio you can gate on — flag low-confidence values for review, keep an audit trail, and query stored results server-side without re-running OCR.

How to ask the OCR API for invoice line items as a structured array, the exact JSON you get back (each cell with its own bbox, vertices and match_ratio), and how to parse it in code and unfold it to CSV.

A scanned PDF is a picture of a table. Skip the retyping and the copy-paste: read each page into structured fields and write it out to a CSV (UTF-8 BOM) that opens straight in Excel. Here's the real-world workflow.