Asgard

Code to support the Asgard low powered computer vision cluster at Loyola University Chicago

Table of Contents

• Asgard
  • Table of Contents
  • About
  • Setup
  • Frontend
  • Backend
  • Intra-Node Communication
  • Architecture Diagrams

About

Asgard is a simple cluster of NVIDIA Jetson Nano 2GB devices that aims to provide a test bed for running and testing Computer Vision models for low powered computing devices.

This repository contains all of the information regarding:

• Setting up a NVIDIA Jetson 2GB for this project
• The frontend web interface for uploading jobs (.img files) to the head node of the cluster
• The backend server code to handle file transmission and storage
• Scripts to send files from the head node to a node
• Architecture diagrams to explain how this cluster is organized
• File format documentation to explain what is an isn't a valid file format

Setup

See JETSON_SETUP.md for more information.

Frontend

Frontend code can be found here.

The frontend website is written in a single HTML, CSS, and JS file. It uses Pico.css as its CSS framework to simplify development.

Backend

The server backend for the website is handled via Flask and Python 3.10+.

Backend code can be found here.

Intra-Node Communication

Architecture Diagrams