Google's OCR Engine, Dressed for the App Store

What it does

This repository packages Google’s Tesseract OCR engine—along with Leptonica, libtiff, libpng, and libjpeg—into a drop-in iOS framework. It handles the cross-compilation for armv7s and arm64, bitcode included, exposing the engine to Objective-C or Swift projects running iOS 9.0 or later.

The interesting bit

This is largely a packaging effort: the author cross-compiled Tesseract and its C dependencies for Apple’s armv7s and arm64 targets so iOS developers don’t have to wrestle with configure scripts and Xcode build settings.

Key highlights

• Bundles Tesseract 3.03-rc1, Leptonica 1.72, and image I/O libraries into one framework
• Ships with bitcode enabled for all bundled libraries
• Supports iOS 9.0+ in both Objective-C and Swift
• Distributed via Carthage (and CocoaPods, per the repo shields)
• MIT licensed wrapper around Apache 2.0 upstream code

Caveats

• The bundled Tesseract 3.03-rc1 is an older release candidate; verify it meets your accuracy needs
• Requires language training data to live in a specific tessdata folder at runtime
• macOS support is only available through a community fork, not this main repo

Verdict

Handy if you need offline text recognition in an iOS app without managing C library builds yourself. Not the right choice if you need macOS support from the main repo or dislike managing a tessdata folder.