Cloud Deployment Guide

This directory contains the cloud deployment configuration for the Face Anti-Spoofing Detection project.

🔗 Repository Structure

• Main Project: spoof-object-detect (this repository)
• Cloud Deployment: cloud-deploy_spoof-detect (separate repository)
  • GitHub: https://github.com/EASONTAN03/cloud-deploy_spoof-detect
  • Render: Auto-deployed from the cloud deployment repo

📁 Current Setup

cloud-deploy_spoof-detect/
├── Dockerfile              # Docker configuration
├── render.yaml             # Render deployment config
├── requirements.txt        # Python dependencies
├── app/
│   ├── main.py            # FastAPI application
│   └── model/
│       └── best.pt        # Trained model (copy from main project)
└── DEPLOYMENT_GUIDE.md    # This file

🔄 Updating Model

When you update the model in the main project:

1. Copy the new model from main project:

   # From the main project directory
   cp outputs/models/Spoof-Detect-1-1/weights/best.pt deploy_models/cloud-deploy_spoof-detect/app/model/

2. Update the cloud deployment repo:

   cd deploy_models/cloud-deploy_spoof-detect
   git add app/model/best.pt
   git commit -m "Update model to latest version"
   git push origin main

3. Render will auto-deploy the updated model

🔧 Configuration

Model Path

The FastAPI app loads the model from app/model/best.pt. Make sure this path is correct.

Environment Variables

Add these to your Render environment if needed:

• MODEL_PATH: Path to the model file (default: app/model/best.pt)
• CONFIDENCE_THRESHOLD: Detection confidence (default: 0.45)
• IOU_THRESHOLD: IoU threshold (default: 0.8)

🚀 Deployment

This deployment is configured for Render.com:

1. Repository: https://github.com/EASONTAN03/cloud-deploy_spoof-detect
2. Auto-deploy: Enabled
3. Environment: Docker
4. Build Command: docker build -t spoof-detect-api .
5. Start Command: uvicorn app.main:app --host 0.0.0.0 --port 8000

📡 API Endpoints

• GET /: Health check
• POST /predict/: Image prediction
  • Input: Image file (multipart/form-data)
  • Output: JSON with predictions

🔗 Integration with Main Project

The main project (spoof-object-detect) contains:

• Training code and notebooks
• Dataset processing utilities
• Model training scripts
• Local inference tools

This cloud deployment provides:

• REST API for predictions
• Docker containerization
• Cloud hosting on Render

📝 Usage Example

import requests

# API endpoint
url = "https://your-render-app.onrender.com/predict/"

# Upload image
with open("test.jpg", "rb") as f:
    files = {"file": f}
    response = requests.post(url, files=files)

# Get predictions
predictions = response.json()
print(predictions)

🔄 Workflow

1. Develop in main project (spoof-object-detect)
2. Train new models in main project
3. Copy best model to cloud deployment
4. Push cloud deployment changes
5. Auto-deploy on Render

📞 Support

For issues with:

• Main project: Create issue in spoof-object-detect repo
• Cloud deployment: Create issue in cloud-deploy_spoof-detect repo
• Render deployment: Check Render dashboard logs