advanced_business_card_reader #

Business/visiting card reader plugin for Flutter.

• Native OCR with Google ML Kit Text Recognition v2 (Android/iOS)
• Native Entity Extraction (beta) for extra phone/email/url signals
• Dart-side fallback parsing for name/company/phones/emails/websites

What changed #

This plugin now uses direct native Google ML Kit through a Flutter MethodChannel:

• Channel: advanced_business_card_reader
• Methods:
  • ocrFromFile
  • extractEntities

The public Dart API remains the same:

• BusinessCardReader.scanFromFile(...)
• BusinessCardReader.scanAutoFromFile(...)
• BusinessCardScanResult / BusinessCardData
• OcrScript values (latin, devanagari, chinese, japanese, korean)

Additive API for double-side cards:

• BusinessCardReader.scanFromFilesFrontBack(...)
• BusinessCardReader.scanAutoFromFilesFrontBack(...)
• BusinessCardSidesResult (front, back, merged)

Install #

dependencies:
  advanced_business_card_reader: ^1.1.1

Platform requirements #

Android #

• Minimum SDK: 26 (required because of ML Kit Entity Extraction)
• ML Kit dependencies are provided by the plugin:
  • com.google.mlkit:text-recognition:16.0.1
  • com.google.mlkit:text-recognition-devanagari:16.0.1
  • com.google.mlkit:text-recognition-chinese:16.0.1
  • com.google.mlkit:text-recognition-japanese:16.0.1
  • com.google.mlkit:text-recognition-korean:16.0.1
  • com.google.mlkit:entity-extraction:16.0.0-beta6

If you want a smaller binary, you can fork and remove script-specific recognizers you do not need.

iOS #

• iOS deployment target: 15.5
• 64-bit devices only (Apple requirement)
• Pods used by this plugin:
  • GoogleMLKit/TextRecognition ~> 9.0.0
  • GoogleMLKit/TextRecognitionChinese ~> 9.0.0
  • GoogleMLKit/TextRecognitionDevanagari ~> 9.0.0
  • GoogleMLKit/TextRecognitionJapanese ~> 9.0.0
  • GoogleMLKit/TextRecognitionKorean ~> 9.0.0
  • GoogleMLKit/EntityExtraction ~> 9.0.0

iOS simulator note #

Google ML Kit currently excludes arm64 for iOS simulator builds. iOS 26+ simulator runtimes are arm64-only, so use:

• a physical iOS device (recommended), or
• an iOS simulator runtime that still supports x86_64 (if available in your Xcode setup).

Entity Extraction models are downloaded on-device with downloadModelIfNeeded() before annotation.

Usage #

import 'package:advanced_business_card_reader/advanced_business_card_reader.dart';

final result = await BusinessCardReader.scanFromFile(
  filePath,
  script: OcrScript.latin,
  defaultIsoCountry: 'IN',
);

print(result.data.name);
print(result.data.company);
print(result.data.phones);
print(result.data.emails);
print(result.data.websites);
print(result.data.rawText);

Auto script:

final result = await BusinessCardReader.scanAutoFromFile(
  filePath,
  preferredScripts: const [
    OcrScript.latin,
    OcrScript.devanagari,
    OcrScript.chinese,
    OcrScript.japanese,
    OcrScript.korean,
  ],
  defaultIsoCountry: 'IN',
);

Double-side card:

final sides = await BusinessCardReader.scanAutoFromFilesFrontBack(
  frontPath: frontImagePath,
  backPath: backImagePath,
  preferredScripts: const [
    OcrScript.latin,
    OcrScript.devanagari,
    OcrScript.chinese,
    OcrScript.japanese,
    OcrScript.korean,
  ],
  defaultIsoCountry: 'IN',
);

print('Front: ${sides.front.data.toJson()}');
print('Back: ${sides.back.data.toJson()}');
print('Merged: ${sides.merged.data.toJson()}');

Parsing behavior #

• OCR text is always parsed by regex + phone_numbers_parser fallback.
• Entity Extraction is best-effort enhancement only.
• phones, emails, and websites are merged as a union of:
  • parser results
  • extracted entities
• If entity extraction fails/model download fails, parser fallback output is still returned.

Unsupported platforms #

Web/desktop are not supported. Calls throw UnsupportedError with a clear message.

Example app #

The example app includes:

• Camera
• Gallery
• File picker
• Script selector (Auto + individual scripts)
• Single-side and double-side scan modes

Run:

cd example
flutter run

License #

MIT