README.md

metadata

title: AI vs Real Image Detection
emoji: 🔍
colorFrom: blue
colorTo: purple
sdk: docker
app_port: 7860
pinned: false
license: mit
short_description: FastAPI ensemble AI image detection service

🔍 AI vs Real Image Detection

A FastAPI-based web service that detects whether images are AI-generated or real using ensemble model prediction.

Features

• FastAPI Backend: High-performance async API
• Ensemble Detection: Uses multiple AI detection models for higher accuracy
• CPU Optimized: Runs efficiently on Hugging Face Spaces free tier
• REST API: Simple POST endpoint for image analysis

API Endpoints

• GET / - Root endpoint with service information
• GET /health - Health check endpoint
• POST /analyze - Upload image for AI vs Real detection
• GET /docs - Interactive Swagger API documentation

Usage

1. Health Check: GET /health
2. Analyze Image: POST /analyze with image file in form data
3. API Docs: Visit /docs for interactive documentation

Response Format

{
  "real_prob": 0.85,
  "ai_prob": 0.15,
  "analysis_time": 2.3,
  "device": "CPU"
}

Models Used

• dima806/deepfake_vs_real_image_detection
• umm-maybe/AI-image-detector
• Organika/sdxl-detector

Built with FastAPI, PyTorch, and Transformers. Optimized for CPU deployment.