CamDoc: Building a Document Scanner App in React Native

Why we built a document scanner app

CamDoc is one of our own products. Most people still need to send paper - a signed contract, a receipt for expenses, a form for a landlord or a school - and most of them no longer own a scanner. What they have is a phone camera, and a photo of a page is a poor substitute for a scan: it arrives skewed, shadowed and unevenly lit, and it is the wrong file type for whoever asked for it.

The goal was narrow on purpose: turn photos of paper into clean, readable, shareable PDFs in a couple of taps, and get the file wherever it needs to go without a save-and-find detour. The App Store title, "JPG to PDF Converter", describes that job in the words people actually search for.

What the app does

• Scans documents and photos to PDF, JPG or TXT.
• Combines multiple pages into one document, with a review step to reorder, rotate or remove pages before export.
• Extracts text from scans with OCR (optical character recognition).
• Lets users create an e-signature once and place it on documents.
• Cleans scans with filters and noise removal so text reads as scanned, not photographed.
• Shares to email, print, and cloud storage including Dropbox, Google Drive, Evernote and OneDrive.

It runs on iPhone and iPad, on Apple-silicon Macs and on Apple Vision Pro, and on Android through Google Play - all from one React Native codebase.

How a phone photo becomes a clean PDF

The core of any document scanner app is the capture pipeline. Every page goes through the same three steps before it is added to a document.

1. Edge detection

The app finds the four corners of the page in the photo. Paper is rarely photographed square-on, and the background - a desk, a lap, a car seat - varies wildly, so the detection has to cope with low contrast between page and surface. When it is unsure, the user can drag the corners rather than retake the photo.

2. Perspective correction

Once the corners are known, a perspective transform maps the skewed quadrilateral back to a flat rectangle. This is what makes a page photographed at an angle look as though it went through a flatbed scanner, and it is the single biggest difference between a scan and a photo.

3. Contrast clean-up and noise removal

Phone photos carry shadows, colour casts and sensor noise. A clean-up pass evens out the lighting and lifts text off the background, and optional filters let users choose between a faithful colour copy and a crisp black-and-white document. The trade-off is file size against readability: the export has to be small enough to email and still legible when printed.

Multi-page documents without retakes

Real documents are captured out of order, sideways, or twice. Instead of forcing a retake, every page lands in a review screen where it can be reordered, rotated or removed before export, so the finished PDF is right first time.

Long documents are also where memory becomes a concern. Each page is a high-resolution image, and holding a dozen of them uncompressed is enough to slow down an older Android phone. Pages are processed and stored one at a time rather than kept in memory as a batch.

OCR, e-signatures and sharing

A scan app becomes a daily tool once it can do more than make a picture of a page. OCR turns a scan into text that can be exported as a TXT file, copied or searched. E-signature lets users sign a contract on the phone instead of printing, signing and scanning it again.

Sharing hands off to the platform's native share sheet, so a finished document goes straight to mail, messaging or any installed app, with direct uploads to Dropbox, Google Drive, Evernote and OneDrive, and printing for the cases where paper is still required.

Why React Native for a scanner app

Camera-heavy apps are often assumed to need fully native code. In practice, the performance-critical parts - capture and image processing - run in native modules, while everything around them (the review screen, settings, paywall, sharing flows and localisation) is product UI that React Native handles well. That split let one small team ship and maintain iOS, iPadOS, macOS, visionOS and Android builds from a single TypeScript codebase.

We go through where that line sits, and when it does not hold, in React Native vs native.

Monetisation and localisation

CamDoc is free to download, with two one-off in-app purchases - Unlimited Pro and PDF Premium - instead of a subscription. For a utility people reach for occasionally, a one-time unlock is easier to say yes to than a recurring charge.

The app is localised into 18 languages. Store listings and in-app copy in a user's own language matter more for a utility than for most apps, because the people searching for "PDF scanner" in Spanish, German or Japanese are rarely served by an English-only listing.

What we would do differently

We would ship OCR in the very first release. Text recognition is what turns a scanner from something used once a month into something used every week, and it shapes decisions elsewhere - file formats, export options and how documents are stored - that are cheaper to make early.

Building something similar?

Document scanning, OCR and e-signatures are now expected features in field-service, logistics, finance and HR apps, not just standalone scanners. If your product needs to capture paperwork from a phone, the same pipeline applies.

See how we approach React Native app development.