🎓 Final Year Project – Automated Shared Album

This is the official repository of our Final Year Project – a smart AI-powered mobile app for automatic photo sharing, duplicate & blur filtering, face recognition, highlights generation, and collaborative album management.

It solves the common problem of manual photo sharing after group events by automatically detecting people in photos and sharing them into their personal albums – all while keeping your gallery clean and organized.

---

🚀 Features

👥 1. Smart Face Recognition & Auto-Sharing

• Uses MTCNN for face detection and FaceNet for embeddings.
• Automatically detects registered users in uploaded photos.
• Shares photos to albums of users appearing in the image.
• Ensures privacy by skipping unregistered users.

🧠 2. Blur & Duplicate Image Detection

• Laplacian method for detecting blurry images.
• Perceptual Hashing (pHash) and Hamming Distance for duplicate detection.
• Blurry/duplicate images go into a separate admin review panel.

📁 3. Shared Album Management

• Album creation with Owner|Admin and Viewer roles.
• Invite participants with Firebase push notifications.
• Real-time role update and access control.
• Album collaboration with multiple users.

📷 4. Image Upload & Management

• Upload up to 20 images at once.
• Images stored in AWS S3: album_id/user_id/filename
• Actions supported:
  • View images (all/shared/flagged)
  • Delete (single/multiple)
  • Unflag blurry/duplicate images
  • Update album cover image

🔔 5. Notifications System

• Real-time notifications using Firebase Cloud Messaging (FCM).
• Notifications persisted in PostgreSQL for viewing history.
• Infinite scroll, auto-sync background updates.

🧵 6. Backend Optimizations (AWS Lambda)

• Docker-based Lambda for scalable image processing.
• Key optimizations:
  • DB connection pooling
  • LRU cache for face embeddings
  • ThreadPool for concurrency
  • Blur detection threshold from ENV
  • FAISS indexing for fast similarity search

🖼️ 7. Highlights Generation (AI-powered)

• One photo picked per album using random logic.
• Captions generated using Gemini AI API.
• Images compiled into a downloadable video using FFmpeg.

🧪 8. Face Recognition Experiments

• Multiple thresholds tested (e.g., 0.5–0.7).
• Model fine-tuned using quality image samples.

🔐 9. Authentication & Authorization

• Google OAuth and email/password support.
• Email verification and JWT-based sessions.
• Role-based access for APIs and screens.
• Session expiration detection via Flutter Provider.

📱 10. Flutter Mobile App

• Built in Flutter with clean & modular UI.
• State management using Provider.
• Custom widgets (InputField, GoogleButton, FlushbarHelper, etc.)
• Responsive layout for different screen sizes.
• Image/video picker, success alerts, and offline handling.

---

🛠️ Tech Stack

Layer	Technology Used
Frontend	Flutter (Dart)
Backend	Node.js (Express)
AI / ML	MTCNN, FaceNet, OpenCV
Search	FAISS
Storage	AWS S3
Serverless	AWS Lambda (Docker)
Database	PostgreSQL (AWS RDS)
Messaging	Firebase FCM
Queue	Bull with Redis
DevOps	Docker, EC2
Security	JWT, Email Verification

---

📊 Project Status

✔ Capstone-I: Architecture, Design, Research
✔ Capstone-II: Full Implementation, Optimizations, and Live Demo Ready
📆 Final Defense Presentation – 31 July 2025

---

📂 Folder Structure (Sample)

/
├── backend/
│   ├── controllers/
│   ├── models/
│   ├── routes/
│   ├── services/
│   └── lambda/ (excluded from GitHub)
├── frontend/
│   ├── lib/
│   │   ├── screens/
│   │   ├── widgets/
│   │   ├── providers/
│   │   └── services/
├── README.md

---

📚 How to Run

Backend (Node.js)

cd backend
npm install
node index.js

Frontend (Flutter)

cd frontend
flutter pub get
flutter run

---

🔒 Environment & Lambda Code

This repository does not include the AWS Lambda source code or .env files for security reasons.

• Lambda functions are deployed using Docker on AWS.
• Sensitive credentials (DB, AWS, FCM keys) are excluded.