🚀 Sentinel AI – Backend

Backend service for a Deepfake Detection Platform that allows users to analyze images/videos and determine whether they are AI-generated or real.

---

🧠 Features

• 🔐 User Authentication (JWT + API Key)
• 🔑 Per-user API Key system
• 🔄 API Key regeneration
• 📡 Deepfake detection (URL & file upload)
• 📜 Scan history tracking
• 🗄️ PostgreSQL (Supabase)
• ⚡ High-performance APIs with FastAPI

---

🏗️ Tech Stack

• Framework: FastAPI
• Database: PostgreSQL (Supabase)
• ORM: SQLAlchemy
• Auth: JWT (python-jose)
• Hashing: Passlib (bcrypt)
• Containerization: Docker

---

📁 Project Structure

.
├── app/
│   ├── __pycache__/
│   │
│   ├── detectors/
│   │   ├── image_model.py
│   │   └── video_detector.py
│   │
│   ├── models/
│   │   ├── image_detect/
│   │   │   └── xception_deepfake_base.keras
│   │   └── video-detect/
│   │       ├── model.ipynb
│   │       └── video_model.keras
│   │
│   ├── routes/
│   │   ├── __pycache__/
│   │   ├── api.py
│   │   └── auth.py
│   │
│   ├── services/
│   │   ├── __pycache__/
│   │   ├── image_scraper.py
│   │   └── ytdlp_service.py
│   │
│   ├── auth.py
│   ├── db.py
│   ├── image_model.py
│   ├── main.py
│   ├── models.py
│   ├── schemas.py
│   └── utils.py
│
├── venv/
├── .env
├── .gitignore
├── cookies.txt
├── LICENSE
└── README.md

---

🐳 Docker Setup (Recommended)

Run the backend without installing Python or dependencies.

---

✅ Prerequisites

• Install Docker Desktop

---

🚀 Getting Started

1️⃣ Clone the repository

git clone <your-repo-url>
cd deepfake-backend

---

2️⃣ Create .env file

DATABASE_URL=postgresql://username:password@host:port/database
use ipv4 instead of ipv6 in case of supabase

⚠️ Use Supabase Session Pooler if required.

---

3️⃣ Run the backend

docker compose up --build

---

4️⃣ Access the API

• Backend: 👉 http://localhost:8080

• Swagger Docs: 👉 http://localhost:8080/docs

---

🛑 Stop the server

docker compose down

---

🔄 Rebuild after changes

docker compose up --build

---

⚠️ Notes

• First build may take time (Docker image build)
• Ensure port 8080 is free
• .env file is required
• No need to install Python locally

---

🔐 Authentication

The API supports two authentication methods:

1️⃣ JWT Token (Recommended)

Use the token received after login:

Authorization: Bearer YOUR_JWT_TOKEN

---

2️⃣ API Key

Use the API key received during registration/login:

x-api-key: YOUR_API_KEY

---

✅ You can use either JWT OR API Key to access protected endpoints.

---

🔑 Authentication Flow

Register

• Creates a new user
• Returns API key

---

Login

• Returns:

  • JWT token
  • API key

---

Regenerate API Key

• Generates a new API key
• Invalidates the old one

---

📡 API Endpoints

Auth

• POST /auth/register → Register user
• POST /auth/login → Login & get JWT + API key
• POST /auth/regenerate-key → Generate new API key

---

Core API

• POST /analyse/url → Analyze media from URL
• POST /analyse/analyse_upload → Analyze uploaded file
• GET /history → Get user scan history

---

🧪 Example Requests

Using JWT

POST /analyse/url

Headers:
  Authorization: Bearer YOUR_JWT_TOKEN

Body:
{
  "url": "https://example.com/video.mp4"
}

---

Using API Key

POST /analyse/url

Headers:
  x-api-key: YOUR_API_KEY

Body:
{
  "url": "https://example.com/video.mp4"
}

---

⚡ Notes on Usage

• Swagger UI supports authentication via the 🔒 button
• JWT is recommended for frontend apps
• API keys are useful for scripts and external integrations

---

🧠 Future Improvements

• 🤖 Model improvements
• ⏱️ Rate limiting (Redis)
• 🌐 Chrome extension
• 📊 Dashboard
• 🔐 Multiple API keys

---

💀 Notes

• bcrypt limit: 72 characters
• Use URL-safe DB credentials
• Prefer Supabase Session Pooler for production

---

👨‍💻 Author

Built as part of a hackathon project — Sentinel AI

---

⭐ Contribute

Pull requests are welcome. Open an issue for major changes.

---

If you want next-level polish, I can:

• add badges (Docker, FastAPI, stars, etc.)
• or make it look like a top-tier GitHub repo (with screenshots + demo GIF)