AI Verify Snap

🌟 Project Summary

AI Verify Snap is a state-of-the-art deepfake detection and digital forensics platform designed to verify the authenticity of digital media. As visual misinformation becomes increasingly sophisticated, this platform serves as a reliable tool to detect synthetic modifications, deepfakes, and manipulated imagery.

By analyzing forensic artifacts through Error Level Analysis (ELA) and advanced Machine Learning models, the platform identifies hidden pixel-level inconsistencies that are invisible to the naked eye. Users can securely upload images to run deep scans, view detailed visual analysis, and generate comprehensive, downloadable forensic reports.

Built with a focus on speed, accuracy, and a premium user experience, AI Verify Snap empowers individuals and organizations to trust their digital content in an era of AI-generated media.

---

🏗️ System Architecture

AI Verify Snap employs a modern, 3-tier microservice architecture:

                  ┌────────────────────────┐
                  │   Next.js Frontend     │ (Port 3000)
                  │ (React, Tailwind, Auth)│
                  └───────────┬────────────┘
                              │ HTTPS REST
                              ▼
                  ┌────────────────────────┐
                  │  Spring Boot Backend   │ (Port 8080)
                  │ (Java 17, JPA, WebFlux)│
                  └─────┬──────────────┬───┘
      PostgreSQL        │              │ HTTP Multipart
      (Neon/Local)      ▼              ▼
     ┌───────────┐    ┌───┴──────────┐┌┴──────────────────────┐
     │ Database  │    │ Google Auth  ││  FastAPI ML Service  │ (Port 8000)
     │   (DB)    │    │ (Third Party)││(PyTorch, ResNet+ELA) │
     └───────────┘    └──────────────┘└───────────────────────┘

1. Frontend (Next.js 16): A client-facing SPA with a custom visual system built using Tailwind CSS and Radix UI. Leverages NextAuth.js (Google OAuth) for secure user authentication.
2. Backend (Spring Boot 3.2.2): Orchestrates database transactions, maps results, syncs user sessions, and proxies media analysis requests to the ML service.
3. ML Service (FastAPI): Runs dual-stream PyTorch inference (using ResNet18 and ELA CNN streams) with automated JET-colormap error heatmaps and SerpAPI Google Lens visual match fallbacks.

---

🚀 Quick Start (Docker Compose)

The easiest way to run the entire stack locally is by using Docker Compose.

Prerequisites

Ensure you have Docker and Docker Compose installed.

Setup Steps

1. Configure Environment Variables: Create a .env file in the root of the repository matching the parameters of the .env.example files:

   DB_PASSWORD=your_secure_password
   SERP_API_KEY=your_serpapi_key
   NEXT_PUBLIC_CLERK_PUBLISHABLE_KEY=pk_test_...
   CLERK_SECRET_KEY=sk_test_...

2. Start the Infrastructure: Run the following command to build and launch all containers:

   docker-compose up --build

3. Access the Application:

   • Frontend: http://localhost:3000
   • Backend REST API: http://localhost:8080
   • Swagger API Documentation: http://localhost:8080/swagger-ui.html
   • ML Service Layer: http://localhost:8000

---

🛠️ Manual Services Setup

If you prefer to run services individually:

1. ML Detection Layer (Python)

cd ML_Service
python -m venv venv
source venv/bin/activate  # On Windows: .\venv\Scripts\activate
pip install -r requirements.txt
python main.py

• Runs on http://localhost:8000

2. Forensic API Gateway (Spring Boot)

Ensure you have JDK 17 and Maven 3.9+ installed:

cd backend
# Set your environment variables (DB_URL, DB_USERNAME, DB_PASSWORD)
mvn clean spring-boot:run -Dspring-boot.run.profiles=dev

• Runs on http://localhost:8080

3. Forensic Console (Next.js)

Ensure you have Node.js 20+ installed:

cd frontend
npm install
npm run dev

• Runs on http://localhost:3000

---

🛡️ Security Hardening & Best Practices

This repository follows production-grade security practices:

• No Committed Secrets: Environment variables are parameterized via .env files.
• Bounded CORS: Pre-flight checks restrict requests to authenticated clients.
• Hibernate Safe Defaults: Production profile (application-prod.yml) enforces ddl-auto: validate to prevent dynamic schema mutations.
• Safe Deserialization: Uses custom JSON-mapping exceptions instead of raw Stack trace leakage in response bodies.