🔍 Food Scanner MVP

A smart iOS application that helps users make informed dietary choices by analyzing food labels instantly. Using advanced OCR technology and AI-powered analysis, Food Scanner MVP identifies allergens, dietary compliance, and safety concerns with a single photo.

✨ Features

📸 Smart Image Recognition

• Instant OCR: Capture food label photos and automatically extract ingredient text using Vision Framework
• High Accuracy: Utilizes iOS's native Vision framework with language correction for reliable text extraction
• Editable Results: Manual correction or direct text input for ingredients

🤖 AI-Powered Analysis

• Multi-Provider Support: Leverages Google Gemini 2.5 Flash for intelligent ingredient analysis
• Comprehensive Safety Assessment: Identifies:
  • ❌ Pork and alcohol derivatives
  • ⚠️ Common allergens
  • 🚫 Unsafe ingredients based on user preferences
  • ✅ Dietary rule compliance

🎯 Personalized Dietary Rules

• Custom Preferences: Support for multiple dietary rules (Halal, Vegan, Vegetarian, Keto, Sugar-Free, etc.)
• Allergen Management: Create and manage personal allergen lists
• Real-Time Compliance Checking: Instant verification against your dietary constraints

🔐 Privacy & Security

• Firebase Integration: Secure authentication and data handling
• Local Processing: Image processing happens on-device for privacy
• No Cloud Storage: Your dietary preferences stay with you

🏗️ Architecture

Technology Stack

• Language: Swift
• UI Framework: SwiftUI
• OCR: Vision Framework (Apple's native solution)
• AI/ML: Google Gemini 2.5 Flash via Firebase AI Logic
• Backend: Firebase (Authentication & Configuration)
• Storage: UserDefaults (Local preferences)

Key Components

Services

• OCRService: Handles image text recognition with language correction
• GeminiService: AI-powered ingredient analysis and safety assessment
• AIService Protocol: Abstraction for multiple AI providers
• ProviderManager: Manages multiple AI providers with fallback support
• KeywordFallbackService: Backup analysis when primary providers fail

ViewModels

• ScanViewModel: Manages image selection, OCR processing, and analysis state
• SettingsViewModel: Handles user preferences (dietary rules, allergens, AI provider selection)
• ResultsViewModel: Manages analysis results display logic

Views

• ScanView: Main interface for image selection and analysis
• ResultsView: Displays comprehensive safety assessment
• SettingsView: Manage dietary preferences and AI provider
• AboutView: Application information and guidance

🚀 Getting Started

Prerequisites

• iOS 14.0 or later
• Xcode 15.0+
• CocoaPods (for dependency management)

Installation

1. Clone the Repository

   git clone <repository-url>
   cd "Food Scanner MVP"

2. Install Dependencies

   pod install

3. Firebase Setup

   • Download GoogleService-Info.plist from Firebase Console
   • Add it to the Xcode project (replace existing file)
   • Ensure Firebase AI Logic is configured

4. Build & Run

   • Open Food Scanner MVP.xcworkspace in Xcode
   • Select your target device or simulator
   • Press Cmd + R to build and run

📖 How to Use

1. Open the App → Navigate to the Scan tab
2. Select Image → Tap "Select Image" to choose a food label photo
3. Review Text → OCR automatically extracts ingredients (editable)
4. Set Preferences → Configure your dietary rule and allergens in settings
5. Run Analysis → Tap "Run Analysis" to check safety
6. Review Results → See detailed findings with specific warnings and notes

Example Scenarios

• Halal User: Detects pork, alcohol, and non-halal ingredients
• Vegan User: Identifies animal products (dairy, eggs, gelatin, honey)
• Allergy Sufferer: Flags specific allergens and cross-contamination risks
• Health-Conscious: Checks for sugar, additives, or other restrictions

🔍 Analysis Output

Each analysis returns:

{
  "containsPork": boolean,
  "containsAlcohol": boolean,
  "allergens": ["string"],
  "unsafeIngredients": ["string"],
  "analysisNotes": "string",
  "isSafeForUser": boolean
}

🛡️ Safety & Privacy

• On-Device Processing: Images are processed locally; only ingredients text is sent to AI
• User Control: You control which AI provider to use
• No Tracking: Application respects user privacy
• Fallback System: If primary AI fails, keyword-based fallback ensures functionality

🐛 Known Issues

• OCR accuracy varies based on image quality and label clarity
• Complex ingredient lists may require manual correction
• Language support currently limited to English

🙏 Acknowledgments

• Google Gemini team for AI analysis capabilities
• Apple Vision Framework documentation
• Firebase team for backend infrastructure
• The open-source community for inspiration and support