Passport Auto-Capture

The SDK finds the passport data page, the page with the photo and the machine-readable zone (MRZ), in the live camera feed, captures a sharp, flattened image and sends it to the Passport OCR API: 0.75 IC per page.

Call

Flutter

import 'package:iapp_ekyc_sdk/iapp_ekyc_sdk.dart';

final client = IappEkycClient(apiKey: 'YOUR_API_KEY');

final result = await DocumentCaptureView.start(
  context,
  client: client,
  documentType: DocumentType.passport,
  locale: EkycLocale.en,
);

Web

import { IappEkyc } from '@iapp-technology/ekyc-sdk';

const ekyc = new IappEkyc({ apiKey: 'YOUR_API_KEY' });

const result = await ekyc.captureDocument({
  mount: document.getElementById('ekyc-mount'),
  documentType: 'passport',
  locale: 'en',
});

iOS

// iOS (Swift; Objective-C also supported)
let config = IappEkycConfig(apiKey: "YOUR_API_KEY", flow: .documentCapture)
config.documentType = .passport
IappEkycSdk.present(from: self, config: config) { result in /* result.document?.rawJSON */ }

Android

// Android (Kotlin; Java also supported)
ekyc.launch(IappEkycRequest.DocumentCapture(config, EkycDocumentType.PASSPORT))

React Native

// React Native
<IappEkycFlow flow="documentCapture" documentType="passport" apiKey="YOUR_API_KEY"
  onResult={handleResult} onError={handleError} onCancel={close} />

Install: Getting started for Flutter and Web, iOS, Android & React Native for the native wrappers and camera permission.

Options

• documentType: passport (DocumentType.passport in Flutter).
• locale: the on-screen language, en, th or zh (EkycLocale.en in Flutter).

Result

The result has typed fields plus the raw OCR response in result.raw (result.document?.rawJSON on iOS), including the parsed MRZ fields. Field list: Passport OCR API.

How auto-capture works

• Boundary: a four-sided shape in the on-screen guide matching the ID-3 passport data-page aspect ratio of 1.42 (125 × 88 mm), within ±0.25.
• Stability: the page must stay steady across recent frames, its corners drifting less than 2% of the frame diagonal.
• Sharpness: blurry frames are rejected before any credit is spent, which keeps the fine MRZ print free of motion blur.
• Perspective correction: the corners are warped to a flat page image of about 300 DPI (1476 × 1039 px) before upload.
• Manual fallback: a shutter button appears after 10 seconds if auto-capture has not fired.

The full algorithm specification is in the SDK source.