Mastering Google Coral Dev Boards for Beginners

Name: Mastering Google Coral Dev Boards for Beginners
Rating: 4.6 (4 reviews)

Machine Learning at the Edge: inference, object detection and pose estimation

Created byDr. Sebastian Marciniak

Last updated 6/2024

English

What you'll learn

Overview of the hardware specifications, capabilities, and ideal use cases for Coral products
Participants will learn how to set up the development environment, configure network settings, and ensure the system is ready for deploying AI models
Participants will learn how to integrate the Edge TPU into applications for various domains such as surveillance, retail, healthcare, or smart home systems
Students will develop skills in implementing AI models for image classification and object detection in Machine Learning based projects

Course content

3 sections • 18 lectures • 1h 0m total length

Introduction0:38
Accelerated machine learning enables faster and more efficient processing of AI tasks for various applications, such as:
- image recognition
- object detection
- natural language processing
and others without having to rely on remote servers.
The Coral Dev Board is particularly useful for applications where low latency and privacy are especially important, such as in robotics, IoT devices, smart cameras and industrial equipment. Developers, such as you, can use the Coral Dev Board to prototype and deploy AI models efficiently.

Coral Edge TPU Board: overview of the hardware specifications and capabilities2:43
Get to know your Coral Edge TPU Dev Board! You will learn about the various ports and features that distinguish this board from many others.
Adding external, durable case0:56
I came across this appealing and sturdy anodized aluminum case with a plexiglass top. It tuned out to be very handy after all.
Make your board alive: installing Mendel Linux on your Coral Edge TPU Dev Board8:43
Installing Mendel Linux requires lots of patience... In this episode, we will do it together. In a few moments, your board will be up and running.
Communicating and setting-up Wi-Fi with the Coral Edge TPU3:09
How do I communicate with my board? How do I connect to Wi-Fi? These are the first two questions that come to your mind at this stage of setting-up your new purchase. Worry-not, here we are going to do all of that together.
Mendel Development Tool (MDT) is an essential tool for developers working with Coral devices, as it streamlines the development and deployment process for edge AI applications, making it more efficient to prototype and deploy machine learning models on Coral hardware.
Bear with me on the journey to fun!
Video inference: locally or over the network3:49
Let us perform inference on TPU vs. CPU as see the results in real time locally vs. over the network.
Image classification and short inference times2:42
Let us classify an image within a few milliseconds with Edge TPU
Sending files and directories to your Coral Edge TPU2:55
When performing inference on images or video offline, you will have to send files to the Coral Edge TPU. In this part you will learn about the "push" command from the Mendel Development Tool set.
Pulling files and directories from Coral to your local machine1:21
If you are ever in the need of editing existing python scripts, you will need to "pull" them from the Coral Dev Board first. The same applies to processed image and video files with annotations.
Detecting objects in an image 12:29
Google provided several images and python scripts to play with. In this image, Coral Edge TPU recognizes a tie and a person.
Detecting objects in an image 23:50
In this example, we will detect people and... kites in an image as well as explore the list of available "coco-dataset labels".
The python script that lets you do all the magic3:04
Learn to change the detection threshold and other parameters though python scripting
Adding eyes to your Coral Edge TPU Board2:49
Why not add vision capabilities to your Coral Edge TPU Dev Board? You can do so much more in real time. Imagination is your limit.

Overview of the hardware specifications2:48
Meet the tiniest member of the Coral family: the Micro development board, along with its Wi-Fi add-on board
Setting-up the ecosystem, booting and downloading examples from Google5:33
Getting started with the Micro is a pure pleasure. Download the example sketches by Google and play in no time at all...
Face detection app running reliably on Edge TPU5:19
Try the face detection example from Google website
Multi-core model cascading6:09
A unique feature of the Dev Board Micro is its ability to transition between models running on the M4 core and models running on the M7 core and Edge TPU. This is called "multi-core model cascading" because information cascades between models on each MCU core.
Time to say "see you later!"1:17
As we wrap up this tutorial, we've delved into the exciting world of edge AI and witnessed the power of these compact yet mighty devices. From performing lightning-fast inference to pose detection, and even pushing the boundaries with multi-core model cascading on the Coral Dev Board Micro, we've seen firsthand the potential they hold for diverse applications.
But remember, this is just the beginning of your exploration. With the knowledge gained here, you're equipped to embark on your own projects, whether it's enhancing surveillance systems, building intelligent IoT devices, or revolutionizing healthcare applications.
The possibilities are truly limitless.
Remember to keep experimenting, keep innovating, and most importantly, keep learning. Your creativity combined with the power of Google Coral Dev Boards holds the potential to shape the future of AI at the edge.
Thank you for joining me on this journey, and until next time, happy coding!

Requirements

Basic experience with Linux
Basic experience with Python is recommended
Machine running Ubuntu Focal Fossa 20.04 LTS (tested to work correctly with this version)

Description

Welcome to "Mastering Google Coral Dev Boards," your comprehensive guide to unlocking the full potential of Google's Coral Dev Boards. This course is designed to empower you with the knowledge and skills needed to harness the power of these cutting-edge devices.

You'll start by getting acquainted with the Google Coral ecosystem and understanding the hardware specifications of both the regular and micro dev boards. From there, we'll dive into the fundamentals of edge AI, exploring topics such as TensorFlow Lite, model optimization, and deployment strategies tailored for the Coral Dev Boards.

As we progress, you'll learn how to perform inference tasks efficiently using pre-trained models and explore techniques for custom model development and training. You'll master pose detection using the Coral Dev Boards, unlocking the ability to build intelligent applications capable of recognizing human poses in real-time.

But the learning doesn't stop there. We'll push the boundaries further by delving into multi-core model cascading on the Coral Dev Board Micro, enabling you to harness the full computational power of these devices for complex AI tasks.

Throughout the course, you'll follow hands-on tutorials, guiding you through practical exercises and projects that reinforce your understanding and skills. By the end, you'll be equipped to tackle a wide range of AI projects at the edge, from smart surveillance systems to innovative IoT applications and beyond.

Join us on this journey to master the Google Coral Dev Boards and unlock the potential of edge AI. Enroll now and start building the future today!

Who this course is for:

This course is designed for beginners seeking to harness the power of Coral Edge TPU and Coral Dev Board Micro for edge AI applications, with a focus on image classification, object detection, and deploying AI models on Mendel Linux

Mastering Google Coral Dev Boards for Beginners

What you'll learn

Explore related topics

Course content

Introduction1 lecture • 1min

Coral Edge TPU Dev Board12 lectures • 39min

Coral Edge TPU Dev Board Micro5 lectures • 21min

Requirements

Description

Who this course is for: