1 of 43

AI6 Lagos: Edge Computing Workshop

2 of 43

George Igwegbe is the Ambassador at Edge Impulse Network. He has experience in designing IoT hardware systems in metering (electricity), Computer Vision. He’s also worked in Hardware and Machine Learning.

He received his Bachelor’s degree in Mechanical Engineering from UNILAG. He is a certified Tensorflow Developer and co-organizer of TinyML Nigeria. Curator of “tinyml-papers-and-projects”.

George is interested in Machine Learning on Embedded Systems and Video Analytics.

George Igwegbe

ML Engineer / Co-organizer AI6 Lagos

Bio:

3 of 43

What you will learn by the end of the workshop

  • Train and deploy model on Jetson Nano
  • Linux basic command
  • Working with Linux
  • Jupyter Notebook / Jupyter Lab
  • Docker
  • SSH / VNC
  • Github
  • Train and deploy model on Edge Devices using Edge Impulse
  • Using the above together!!!

4 of 43

WEEK 1: INTRO AND SETTING UP

5 of 43

5

What are Edge Devices: are pieces of equipment that serve to transmit data between the local network and the cloud. An edge device works by being near the source of the data it manages.

Wide Range of “Edge” Applications

Jetson Nano

6 of 43

6

What are Edge Devices: are devices whose compute, memory, and energy resources are constrained and cannot be easily increased or decreased. These constraints may be due to form-factor considerations and cost.

Wide Range of “Edge” Applications

Jetson Nano

7 of 43

What is Edge Computing ?

Edge Devices possess both communication and computational capabilities.

Edge Computing - refer to computations being performed as close to data source as possible, instead of on far-off remote location. Edge computing is the practice of processing data physically closer to its source.

2020 -> 25 Billion devices

2025 -> $3.9 - $11 Trillion p.a

8 of 43

Overview of Artificial Intelligence

AI, in the broadest sense, describes the different ways a machine interacts with the world around it. To maximize our chance of achieving a given goal. At it core, ML is a simply way of achieving AI.

9 of 43

What is Machine Learning ?

Machine learning uses data and produces a program to perform a task

10 of 43

Introduction to AI

  • ML - subset of AI
  • Machines learn to do task without explicitly programmed to do so.
  • Reinforcement learning, decision tree,DL,clustering…..
  • DL - subset of ML
  • DL learns to do task without explicitly programmed to do so.
  • Mimics the neurons in a human brain.
  • CNN, RNN, AutoEncoder, Transformer…..

Machine Learning

Deep Learning

11 of 43

Edge AI

GPT3

GPT4

DALL.E

CLIP

PaLM 2

Deep Learning has a size problem?

*Shifting from state-of-the-art accuracy to state-of-the-art efficiency

175 Billion

170 Trillion

< 540 Billion

12 of 43

GPT3, GPT4, CLIP…..

*Shifting from state-of-the-art accuracy to state-of-the-art efficiency

Deep Learning has a size problem?

13 of 43

What is an Edge AI ?

Edge AI is the deployment of AI applications in devices throughout the physical world. It’s called “edge AI” because the AI computation is done near the user at the edge of the network, close to where the data is located, rather than centrally in a cloud computing facility or private data center.

14 of 43

AI on the Cloud vs AI at the Edge

15 of 43

Why is ML Moving to Edge ? Benefit of Edge AI

BVAS?

16 of 43

Frictionless Access Control

Understanding the flow of traffic through intersections to better design roadways and reduce traffic congestion

Traffic Flow Management

Faster boarding for airplanes and trains and determining the occupancy of transportation carriages

Retail and Merchandising

Optimization

Understand customer behavior to provide richer experiences and optimize store layouts

Logistics and Cargo Companies

Keep better track of metrics such as how long it takes to load cargo onto an airplane and identify areas of inefficiency

Making parking management more efficient by minimizing the search for an open spot and more accurately determining the duration that a space is occupied

Examples of Edge Applications

Parking Lot Management

17 of 43

Examples of Edge Applications

18 of 43

Device: Arduino BLE

Flash memory: 1MB

Processor: 64MHz

Weight: 5.0g

Comm: I2C, SPI, UART..

Language: C/C++

Device: OpenMV H7+

Flash memory: 16MB

Processor: 480MHz

Weight: 17.0g

Comm: I2C, SPI, CAN, UART..

Language: C++/Micropython

Concepts in Edge AI

19 of 43

Device: Raspberry Pi

SDRAM: 1,2,4,8GB

Processor: 1.5GHz

Weight: 46.0g

Comm: I2C, SPI, CAN…

Language: Any

Device: Jetson Nano

SDRAM: 2,4GB

GPU: 128-core Maxwell

Processor: 1.43 GHz

Weight: 77.0g

Comm: I2C, SPI, CAN, UART.

Language: Any

Concepts in Edge AI

Device for the workshop

20 of 43

21 of 43

Jetson Nano: Device for the workshop

22 of 43

23 of 43

Nvidia Devices

Jetson Nano: Device for the workshop

Edge Impulse: Training model for the workshop

24 of 43

Resources and Expectation

25 of 43

Workshop Github Repository

Workshop Github - Link

26 of 43

Chapter 5: Tools and Expertise, page 136

27 of 43

What you will build by the end of the workshop

Image Classification

Image Regression

DLI Projects - Link

28 of 43

Deploying model on Mobile Phone/Laptop

Deploying model on Jetson Nano

Using edge platform

What you will build by the end of the workshop

29 of 43

Jetson Playlist - Link

Setting up Jetson Resource

30 of 43

Certificate at the end of the program…after 4 weeks & nurture Jetson Ambassadors in Nigeria..

31 of 43

SETTING UP JETSON

32 of 43

32

33 of 43

34 of 43

35 of 43

36 of 43

37 of 43

Training Models: Setting up Edge Impulse Account

38 of 43

39 of 43

40 of 43

41 of 43

42 of 43

43 of 43

THANK YOU