Receipt ScannerGuides

Scan PDF Receipts & Invoices On-Device

·2 min read

Modern expenses rarely arrive only on thermal paper. Airline tickets, SaaS subscriptions, utility bills, hotel folios, and online marketplace orders usually land in your inbox as PDF invoices.

Photographing a laptop monitor to scan an emailed PDF introduces moiré patterns, glare, and low character contrast. Importing the PDF directly into Receipt Scanner avoids all of those problems while keeping every page 100% on your device.

How on-device PDF scanning works

Most online “PDF to Excel” or “PDF invoice OCR” tools upload your document to a remote server to extract the text. Receipt Scanner processes the file inside your browser tab instead:

  1. Local page rendering (pdf.js): Each PDF page is rendered onto an off-screen canvas in your browser using self-hosted fonts and character maps—no external CDN calls and no server upload.
  2. Contrast & scale normalisation: Pages are rasterised cleanly and resized so small invoice type stays sharp without exhausting mobile memory.
  3. WebAssembly OCR & field parsing: The on-device PP-OCRv6 engine reads the page and extracts the merchant, date, subtotal, tax (including GST/CGST/SGST, VAT, and US sales tax), invoice number, tax ID, currency, and grand total.
  4. Page-by-page review: Each page becomes its own item in your scan queue so you can verify flagged fields before saving to your active expense book.

Batch limits and multi-page PDFs

  • Up to 20 pages per batch: You can drop a single multi-page PDF statement, several one-page PDF invoices, or a mix of photos and PDFs in one batch (up to 20 total receipt pages).
  • Multi-page invoices: Because each page is queued separately, if a two-page invoice puts line items on page 1 and the grand total on page 2, confirm the summary page containing the total and skip the extra page so you do not record a duplicate expense.

When a PDF will not open

A few PDF edge cases are rejected intentionally with a clear message rather than failing silently:

  • Password-protected PDFs: Encrypted bank statements or utility bills must be unlocked first (for example via your system’s “Print to PDF” after opening with the password) before importing.
  • Oversized or corrupted files: Strict local memory and page-size guards prevent a malformed PDF from freezing a mobile browser tab.

Tips for cleaner invoice extraction

  • Prefer the original PDF over a screenshot: Screenshots compress thin fonts and crop margins that help the parser distinguish header metadata from totals. For paper bills, follow how to photograph a receipt for better OCR.
  • Check invoice numbers and tax IDs: On B2B invoices (such as GSTIN or VAT invoices), always verify extracted registration numbers and tax totals against the document before exporting to Excel or CSV.
  • Decide whether to store page images: Just like camera photos, you can toggle Keep receipt photo on the review screen—leave it on when you need visual proof in your local database, or turn it off to keep IndexedDB lean.

Put it into practice

Scan receipt