Arduino in TinyML with Edge Impulse
- 1. Arduino in TinyML Edge Impulse & Nano 33 BLE Sense
- 2. Pre-Requisites Basic ML Background Understanding of ML Concepts like Training/Testing/Quantizations/Neural Networks
- 3. Overview Machine Learning What is Machine Learning , ML vs. Traditional Programming Deep Learning What constitutes deep learning TinyML Introduction to TinyML Why TinyML ? Advantages of TinyML TinyML Applications Where all TinyML is used today How to TinyML ? Techniques used in the Implementation of TinyML Edge Impulse Introduction to Edge Impulse Overview Overview of the platform Boards Different boards that can be integrated into edge impulse Data Collection Data collection process in Edge Impulse Training & Testing Training based on the gathered data and testing for the overall accuracy Deployment Deploying the project to your microcontroller board
- 4. Machine Learning (ML) ML is the study of computer algorithms that can improve automatically through experience and by the use of data. -Wikipedia Machine Learning Data Results Rules Future Predictions
- 5. Deep Learning Source: What are Neural Networks? | IBM
- 6. TinyML TinyML lies at the intersection of the two exciting field of Embedded Systems and Machine Learning opening up the doors to a wide variety of applications.
- 7. Why TinyML ? Anomaly Detection Predictive Maintenance Keyword Spotting Recognizing a particular keyword (Eg. Alexa , Hey Google) Object Detection Searching through images for a particular object Applications Privacy Data need not be transferred for processing Improved Performance Low Power Consumption, Latency & Bandwith Edge Processing Wide range of applications, process at the source Benefits
- 8. Additional TinyML Applications Manufacturing Predictive Maintenance ( Santagostino's predictive maintenance for HVAC uses Nano RP2040 Connect (arduino.cc) ) Healthcare Realtime Monitoring ( Accurate Health Information Accessible to Everyone (ouraring.com) ) Agriculture Crop & Livestock Monitoring (Edge Impulse - Website) Retail Inventory Monitoring , Out of Stock Notifcations (Why TinyML use cases are taking off (techtarget.com))
- 9. How to TinyML? Techniques Pruning Quantization Quantization Aware Training Post Training Quantization Tools Edge Impulse 1 Tensorflow Lite for Microcontrollers 2 CodeCraft (Seeed Studio) 4 AIfES 3 A review on TinyML: State-of-the-art and prospects - ScienceDirect
- 10. Generic TinyML Pipeline / Workflow 1 Data Collection 2 Data Preprocessing 3 Model Design 4 Model Training 5 Model Evaluation 6 Model Conversion 7 Deployment 8 Inferencing TinyML
- 11. Q&A
- 12. Edge Impulse - Practical Demonstration!
- 13. Initial Board Setup Edge Impulse Firmware Arduino Nano 33 BLE Sense (edgeimpulse.com) Edge Impulse CLI Installation (edgeimpulse.com) Arduino CLI Installation - Arduino CLI
- 14. Data Collection
- 15. Training
- 16. Testing
- 17. Deployment
- 19. Day 1 March 21 Introduction to Arduino Overview of Various Arduino Boards Getting Started with Arduino Day 2 March 22 Averse to writing code? Program Arduino using Blocks Day 3 March 24 Arduino & Industry Day 5 March 26 End to End TinyML application using Sensa & AIFES with Arduino Day 4 March 25 Exploring TinyML with Arduino & Edge Impulse Overview of the Arduino Week Feedback : https://forms.gle/E1kxWcTdaWqEzueRA
Editor's Notes
- #9: Oura, Predictive MT. Manufacturing: TinyML-powered predictive maintenance can reduce the downtime and costs associated with equipment failure. Retail: TinyML can be used to monitor inventories and send alerts. This can prevent out-of-stock situations. Agriculture: TinyML devices can be used to get real-time data by monitoring crops or livestock. Healthcare: Real-time health monitoring enabled by TinyML devices can deliver better and more personalized patient care. - OURA Ring
- #10: Show toolset - Edge Impulse , Tensorflow Lite
- #12: Short Q&A 2-5min
- #19: EON Tuner TFLM AIfES