Offline Facial Attendance with Kiosk

Project Title:
Offline Facial Recognition Module + Mobile App Packaging (iOS & Android) with Kiosk Mode

---

Project Description:
We have already built an application using Base44 (web app / PWA). The current version includes facial recognition powered by the cloud API from Google Cloud Vision. However, this solution no longer meets our needs due to accuracy issues beyond ~100 users and dependency on external services.

We are now looking to implement a fully offline (on-device) facial recognition system, connect the application to an on-premise database, and deploy it as a kiosk-style mobile application (tablet or wall-mounted device) with automatic face detection.

---

Project Objectives:
1. Remove all dependency on external/cloud APIs
2. Implement reliable on-device facial recognition
3. Ensure full offline functionality (no internet required for recognition)
4. Integrate with an on-premise Microsoft SQL Server database
5. Implement a kiosk mode with automatic face detection trigger
6. Build and deliver native mobile apps:
- iOS (App Store)
- Android (Google Play)
7. Assist with publishing using our developer accounts

---

Technical Scope:

### 1. Offline Face Detection & Recognition
The developer should implement:

- On-device face detection (e.g. ML Kit Face Detection or equivalent)
- Facial embedding generation using an embedded model:
- MobileFaceNet / FaceNet / ArcFace (TensorFlow Lite or equivalent)
- Local face matching (cosine similarity or similar method)
- Configurable recognition threshold

---

### 2. Kiosk Mode (Critical Requirement)
The application will run on a tablet or mobile device fixed on a wall.

Expected behavior:
- App runs in full-screen kiosk mode (no user navigation outside the app)
- Camera is continuously active (or intelligently triggered)
- Automatic detection:
- When a person appears in front of the camera → trigger face detection
- When a face is detected → trigger recognition process
- Display result:
- Identified user (name / ID) OR
- “Unknown user”
- Auto-reset after a few seconds for the next person

Additional requirements:
- Prevent user from exiting the app (guided access / kiosk mode Android & iOS)
- Handle continuous usage (memory, battery, performance optimization)
- Smooth UX (no lag, near real-time detection)

---

### 3. Secure Local Storage
- Local database (SQLite / Realm / ObjectBox)
- Store:
- facial embeddings
- user identifiers
- images (optional, preferably encrypted)
- Data encryption recommended

---

### 4. Integration with On-Premise Database
The application must connect to:

- Microsoft SQL Server

Requirements:
- Secure connection via API layer / VPN / gateway (no direct DB exposure)
- Synchronization capabilities:
- User data
- Facial data (optional)
- Offline-first architecture:
- Full local operation without network
- Sync when connection is available
- Conflict resolution strategy

---

### 5. Integration with Existing App (Base44)
- Take over the existing application
- Integrate the native recognition module:
- via Capacitor plugin / native bridge
- OR propose migration to Flutter / React Native if needed
- Adapt UI for kiosk usage

---

### 6. Mobile App Packaging
- Build deliverables:
- Android (APK + AAB)
- iOS (IPA)
- Configure:
- camera permissions
- background execution (if needed)
- Test on real devices (tablet preferred)

---

### 7. App Store Deployment
The developer must:

- Configure developer accounts:
- Apple Developer Program
- Google Play Console
- Generate certificates & provisioning profiles
- Prepare:
- app icons
- splash screens
- store metadata
- Publish apps under our credentials

---

Key Requirements:
- Fully offline recognition capability (mandatory)
- Kiosk mode stability for continuous operation
- Real-time face detection trigger
- Secure sync with MS SQL Server
- Scalable to at least 500–1000 users locally
- Clean, maintainable, and documented code

---

Expected Deliverables:
- Full source code
- Compiled mobile apps (Android & iOS)
- Technical documentation:
- architecture
- installation/setup
- kiosk configuration
- sync mechanism with MS SQL Server
- Store deployment guide
- Successful publication on our accounts

---

Required Skills:
- Computer vision / facial recognition experience
- TensorFlow Lite / embedded ML models
- Mobile development (Android & iOS)
- Experience with kiosk / locked-down apps
- Experience with Microsoft SQL Server integration
- API design for secure connectivity
- Capacitor / Flutter / React Native
- Proven track record (portfolio required)

---

Nice to Have (Bonus):
- Anti-spoofing (real face vs photo detection)
- Performance optimization (<1 second recognition)
- Experience with edge AI / real-time video processing
- Offline-first architecture expertise

---

Please include in your proposal:
- Proposed technical stack
- Approach for kiosk mode implementation
- Approach for MS SQL on-premise integration
- Relevant past experience
- Estimated timeline
- Estimated budget
- Demo / GitHub / previous work

---