Food Calorie Frontend

Flutter frontend for an AI-powered food calorie recognition app. The app lets users pick or capture one food photo or upload a batch of images, sends them to a backend prediction API, and displays predicted food name, calories, confidence, nutrition, ingredients, portion-adjusted calories, healthier alternatives, and recent scan history.

Key Features

• Capture photos using camera or pick from gallery.
• Support single-image and batch uploads with preview before submit.
• Send image or batch data to the backend as multipart upload.
• Show flexible prediction output (food, calories, confidence, nutrition, ingredients).
• Adjust portion multiplier and fetch updated calories from the backend.
• Load healthier alternatives with calorie reduction details.
• Maintain lightweight recent prediction history.
• Built with Riverpod and Material 3.

Tech Stack

• Flutter
• Dart
• Riverpod / Flutter Riverpod
• image_picker
• http

Requirements

• Flutter SDK compatible with Dart 3.10.4
• Dart SDK 3.10.4+
• Running backend API reachable from emulator/device

Setup

1. Install dependencies.

flutter pub get

2. Run the app.

flutter run

Configuration

Backend Base URL

The app resolves backend URL in this order:

1. API_BASE_URL from --dart-define
2. Fallback constant in lib/core/api_service.dart

Run with a custom backend URL:

flutter run --dart-define=API_BASE_URL=http://your-backend-host:8000

Current default fallback URL is set in lib/core/api_service.dart.

Network Tips by Target

• Physical Android device: use your computer LAN IP, and ensure both devices share the same network.
• Android emulator: usually http://10.0.2.2:8000 to reach host machine backend.
• iOS simulator: usually http://127.0.0.1:8000 for local backend.
• Web: ensure backend CORS allows the web origin.

Backend API Contract

The frontend expects these endpoints:

• GET / for connection test
• POST /predict for food classification
• POST /predict/batch, POST /batch-predict, or POST /predict-batch for multiple image predictions
• POST /portion-adjust, POST /adjust-portion, or POST /predict/portion for portion multiplier updates
• POST /healthier-alternatives, POST /alternatives, or POST /food/alternatives for lighter option suggestions

Upload format for POST /predict:

• Content type: multipart/form-data
• Image field name: file

Batch upload field names that the frontend will try:

• files
• images
• file

Minimal response example:

{
  "food": "Pizza",
  "calories": 285,
  "confidence": 0.93,
  "ingredients": ["Dough", "Cheese", "Tomato Sauce"]
}

Batch responses can return a results, items, or predictions array. Ingredient details may include confidence, and portion/alternatives responses can add adjustedCalories, adjustedNutrition, or alternatives data.

Project Structure

• lib/main.dart: App entry point and theming
• lib/app/app_shell.dart: Bottom-nav app shell for scan, diet, history, and profile
• lib/core/api_service.dart: API communication and timeout handling
• lib/features/food_scan/view/screens/food_scan_screen.dart: Food scan screen
• lib/features/food_scan/viewmodel/food_scan_view_model.dart: Food scan state and actions
• lib/features/food_scan/model/food_scan_models.dart: Prediction data models
• lib/features/diet/view/screens/diet_dashboard_screen.dart: Diet dashboard
• lib/features/diet/viewmodel/diet_view_model.dart: Diet state and actions
• lib/features/diet/model/diet_models.dart: Diet and profile models
• lib/features/user_profile/view/screens/profile_screen.dart: Profile setup screen
• lib/features/user_profile/viewmodel/profile_view_model.dart: Profile state and actions
• lib/features/history/view/history_page.dart: History view layer
• lib/features/history/view/scan_detail_page.dart: History detail view

Useful Commands

flutter analyze
flutter test
flutter build apk
flutter clean

Notes

• This project is private (publish_to: none).
• Prediction history is currently in-memory and resets on app restart.