Deploy optimized ML models with ONNX Runtime on Arm platforms
Introduction
Understand ONNX fundamentals and architecture
Set up your development environment
Generate a synthetic Sudoku digit dataset
Train the digit recognizer
Run inference and evaluate the model
Build the Sudoku processor pipeline
Optimize the model for Arm64 deployment
Deploy the model to Android
Next Steps
Deploy optimized ML models with ONNX Runtime on Arm platforms
Introduction
Understand ONNX fundamentals and architecture
Set up your development environment
Generate a synthetic Sudoku digit dataset
Train the digit recognizer
Run inference and evaluate the model
Build the Sudoku processor pipeline
Optimize the model for Arm64 deployment
Deploy the model to Android
Next Steps
Who is this for?
This is an advanced topic for developers who want to build, optimize, and deploy machine learning models using ONNX on Arm64-based platforms such as Raspberry Pi, Arm-based laptops, cloud instances, or Android smartphones.
What will you learn?
Upon completion of this Learning Path, you will be able to:
- Explain what ONNX is and how it enables model portability across ML frameworks
- Build and export a neural network model in Python to ONNX format
- Run inference using ONNX Runtime on Arm64 platforms
- Apply model optimization techniques to improve performance
- Deploy an optimized ONNX model in an Android application
Prerequisites
Before starting, you will need the following:
- A development machine with Python 3.10 or 3.11 installed (Prebuilt ONNX Runtime packages for Arm platforms don’t yet support Python 3.12)
- Basic familiarity with PyTorch or TensorFlow
- An Arm64 device such as a Raspberry Pi or Android smartphone
- Android Studio (required only for the final deployment section)