SG60 Heritage Lens — The Swifties

Discover Singapore’s heritage through your camera.

Discover Singapore’s heritage through your camera. This SwiftUI iOS app uses a Core ML model to recognize landmarks, foods, transport, and cultural icons, then shares bite‑sized history while you earn badges and track progress.

Note: This project was created for a competition and is open to contributions.

---

Table of Contents

• Features at a glance
• Tech stack
• App structure (key files)
• How the scanner works
• Getting started
• Configuration
• Contributing
• Roadmap
• Troubleshooting
• License
• Acknowledgements

Features at a glance

• 📸 Landmark scanner (AI): Take a photo or pick from library; Core ML + Vision classifies the image and shows contextual info.
• 🗺️ Explore and learn: Starter screens for Featured and Nearby exploration flows.
• 🏆 Gamified learning: Badge system across themes (Colonial Heritage, Modern Marvel, Hawker Pro, and more).
• 👤 Profile & progress: Level, stats, and overall completion.
• ⚙️ Settings: Dark mode, language preferences, and notification toggles.

What it does

• Landmark scanner (AI): Take a photo or pick from the library; Core ML + Vision classifies the image and shows contextual info.
• Explore and learn: Starter screens for Featured and Nearby exploration flows.
• Gamified learning: Badge system tracks progress across themes (e.g., Colonial Heritage, Modern Marvel, Hawker Pro).
• Profile & progress: See level, stats, and overall completion.
• Settings: Dark mode, language preferences, and notification toggles.

Tech stack

• SwiftUI for UI and navigation (TabView, NavigationView)
• Vision + Core ML for on‑device image classification (model: Recognition.mlmodel)
• UIKit bridge for image capture/library (UIImagePickerController)
• Supabase (planned/early integration)

App structure (key files)

• AppDelegate.swift — App entry point that hosts ContentView.
• ContentView.swift — Simple demo login screen that navigates to the main tab bar when “authenticated”.
• NavigationTab.swift — Tab bar with Home, Explore, Badges, Profile, and Settings.
• HomepageView.swift — Landing page with quick links and a scanner card.
• CoremlView.swift — Main scanner UI (pick/take photo, run prediction, show result and info text).
• ViewModel.swift — CoremlViewModel that loads the Core ML model and performs predictions via Vision.
• ImagePicker.swift — SwiftUI wrapper around UIImagePickerController.
• BadgesView.swift — Badge grid with per‑badge progress and completion states.
• ProfileView.swift — Basic profile header, level, and navigation to achievements/badges.
• SettingsView.swift — Appearance, language, notifications, and housekeeping actions.
• AccountData.swift — App constants for progress/badges and info snippets for recognized items.
• Supabase.swift — Supabase client initialization (see Security note below).
• Recognition.mlmodel — Core ML classification model packaged in the app.

How the scanner works (high‑level)

1. User selects/takes a photo in CoremlView via ImagePicker.
2. CoremlViewModel resizes the image (360×360), creates a Vision request with Recognition.mlmodel, and runs classification.
3. The top class label is formatted and displayed. CoremlView maps known labels to friendly descriptions.

Architecture (visual)

flowchart TD
	A[SwiftUI Views (CoremlView, Homepage, Badges, Profile, Settings)]
	B[ViewModel (CoremlViewModel)]
	C[Vision VNCoreMLRequest]
	D[Core ML: Recognition.mlmodel]
	E[UIKit Image Picker]
	F[(Supabase Client)]

	A --> B
	B --> C
	C --> D
	B --> E
	B -. planned .-> F

Loading

Getting started

Prerequisites

• macOS with Xcode (recommended Xcode 15 or newer). The project uses modern SwiftUI previews (#Preview) that target recent SDKs.
• iOS Simulator or an iOS device.

Run

1. Clone the repository.
2. Open SG60 Heritage Lens The Swifties.xcodeproj in Xcode.
3. Select an iOS Simulator (or a provisioned device).
4. Build and run.

Demo login

• The current demo uses hard‑coded credentials in ContentView.swift:
  • Email: [email protected]
  • Password: password123
• On success, you’ll be taken to the main tab interface.

Permissions you must add (Info.plist)

If your target doesn’t already include them, add the following usage descriptions for camera/photo access:

• NSCameraUsageDescription — e.g., “Camera access is required to scan landmarks.”
• NSPhotoLibraryUsageDescription — e.g., “Photo library access is required to select images for recognition.”
• (Optional) NSPhotoLibraryAddUsageDescription — if you later save images back to the library.

Key	Why it’s needed
NSCameraUsageDescription	Access device camera to scan landmarks
NSPhotoLibraryUsageDescription	Pick photos from library for recognition
NSPhotoLibraryAddUsageDescription (optional)	Save edited images back to the library

Configuration

Supabase

• File: SG60 Heritage Lens The Swifties/Supabase.swift
• Replace the placeholder URL/key with your own project values from Supabase.
• Security note: Never commit service role keys. Prefer storing secrets outside source control (e.g., .xcconfig files, build settings, or remote configuration). The current repo contains a public anon key for development only.

Authentication

• The app currently uses a demo login in ContentView.swift. For production, integrate Supabase Auth (email/password, OAuth, or OTP) and remove hard‑coded credentials.

Localization

• SettingsView offers language pickers (English/Chinese/Malay/Tamil/Singlish beta). Wire these to a localization system (e.g., .strings files) to translate UI and content.

Contributing

Contributions are welcome!

• Fork the repo and create a feature branch.
• Keep changes focused and include a clear description and screenshots/video when UI changes.
• Where possible, include small unit/UI tests.
• Open a Pull Request.

Good first issues (ideas)

• Replace demo login with Supabase Auth and user profiles.
• Fill out the Explore views (Featured/Nearby) and add map/location support.
• Persist badge progress and user stats.
• Improve result text and add richer content (links, images, “then & now” comparisons).
• Add localization and accessibility passes.
• Move Supabase keys to a safer configuration path.

Roadmap (non‑exhaustive)

• Real auth + profiles synced via Supabase
• Explorer with map and proximity alerts
• Badge achievements, sharing, and leaderboards
• Offline model updates and on‑device improvements
• Full localization (EN/ZH/MS/TA + Singlish easter eggs)

Troubleshooting

• Build fails about permissions: ensure the Info.plist usage keys above are present.
• Camera not working in Simulator: use a real device, or test via photo library.
• Model crashes or no predictions: verify Recognition.mlmodel is included in the target and that the image is resized to 360×360 (as in the code).

License

No explicit license is included. By default, all rights reserved. Consider adding a LICENSE file (e.g., MIT/Apache‑2.0) if you want broader community use.

Acknowledgements

• Thanks to the competition organizers and contributors.
• Built with SwiftUI, Vision, Core ML, and Supabase.