NGP-Server

A server implementation for converting Seeedstudio recamera images into 3D scenes using NVIDIA Instant NeRF.

Features

• Multiple processing pipelines for different use cases:
  • Art & Assets: Optimized for artwork and physical objects
  • Food Scanner: Specialized for food recognition and 3D modeling
  • General Purpose: Standard 3D model generation
• GPU-accelerated processing using NVIDIA Instant NeRF
• RESTful API interface
• Configurable processing settings

Cloud Deployment

Prerequisites

• Google Cloud SDK installed
• Access to Google Cloud with GPU quotas enabled
• Git

Deployment Steps

1. Clone the repository:

git clone https://github.com/Hey-Salad/NGP-Server.git
cd NGP-Server

2. Enable required Google Cloud APIs:

gcloud services enable compute.googleapis.com
gcloud services enable containerregistry.googleapis.com

3. Deploy to Google Cloud:

gcloud app deploy app.yaml

API Documentation

Endpoints

• GET /api/health - Health check
• GET /api/processors - List available processors
• GET /api/settings - Get processing settings
• POST /api/process - Process an image

Processing an Image

curl -X POST http://[YOUR-APP-URL]/api/process \
  -F "file=@image.jpg" \
  -F "type=art" \
  -F 'settings={"resolution": "1024x1024", "style": "realistic"}'

Development

While the application is designed to run on Google Cloud with GPU support, you can run parts of it locally for development:

# Install dependencies
pip install -r requirements.txt

# Run locally (without GPU features)
python app.py

License

MIT License 27/01/2025 SALADHR TECHNOLOGY LTD - PETER MACHONA