MXNet Workshop

Note: This repository is archived and no longer actively maintained.

A comprehensive workshop for learning and implementing deep learning models using MXNet, with examples ranging from basic concepts to deployment on Raspberry Pi and integration with AWS services.

Overview

This repository contains notebooks, scripts, and resources for a hands-on MXNet workshop. It covers everything from basic MXNet API usage to deploying models on edge devices and integrating with cloud services.

Repository Structure

• 00_mxnet_api.ipynb: Introduction to MXNet API
• 01_introduction.ipynb/py: Basic concepts and introduction to MXNet
• 02_mnist/: Training and prediction with MNIST dataset
  • Training models (MLP and LeNet)
  • Prediction examples
  • Dataset download scripts
• 03_piSetup/: Scripts for setting up Raspberry Pi devices
• 04_piModels/: Running models on Raspberry Pi
• 05_piMxnet/: MXNet on Raspberry Pi with camera integration
• 06_piPolly/: Integrating AWS Polly with Pi applications
• 07_piRekognition/: Using AWS Rekognition with Pi applications
• models/: Pre-trained models and download scripts

Getting Started

Prerequisites

• Python 3.x
• MXNet
• Jupyter Notebook
• AWS account (for Polly and Rekognition examples)
• Raspberry Pi (for Pi-specific examples)

Setup

1. Clone this repository:

   git clone https://github.com/yourusername/mxnetworkshop.git
   cd mxnetworkshop
   

2. Download the MNIST dataset and models:

   cd 02_mnist
   ./get_mnist_from_web.sh
   
   cd ../models
   ./get_models_from_web.sh
   

3. For EC2 setup (optional):

   ./ec2_setup.sh
   

4. For Raspberry Pi setup:

   cd 03_piSetup
   ./setup.sh
   

Workshop Modules

1. MXNet Introduction

Learn the basics of MXNet and its API.

2. MNIST Examples

Train and evaluate models on the MNIST dataset:

• Multi-layer Perceptron (MLP)
• LeNet Convolutional Neural Network

3-5. Raspberry Pi Deployment

Set up and run MXNet models on Raspberry Pi devices, including camera integration.

6-7. AWS Integration

Integrate AWS services with your Pi applications:

• AWS Polly for text-to-speech
• AWS Rekognition for image recognition

Cleanup

When finished with EC2 instances:

./ec2_cleanup.sh

License

[Include appropriate license information here]

Acknowledgments

• MXNet team and contributors
• AWS for cloud services integration examples