Discover the Power of Coral: AI Applications at the Edge

Table of Contents

1. Introduction to Coral
2. What is Coral and what can it do?
3. Examples of Coral applications
   • Object detection
   • Object classification
   • Body part segmentation
   • Gesture recognition
   • Human figure detection
   • Facial detection
4. Innovations using Coral
   • Care.ai's self-aware Healthcare environments
   • Olea's smart water meter sensor
   • Hamburgini's AI cooking system
5. New products in Coral
   • M.2 modules
   • Coral Accelerator Module
   • Coral Dev Board Mini
6. Updates in Coral software
7. Coral partnership program
8. Introducing the new Coral Dev Board Micro
9. Conclusion and call to action
10. Resources for learning and using Coral

🚀 Introduction to Coral

Welcome to the Google I/O session of Coral, where we will explore the expanding landscape of Edge AI. Coral is a platform developed by Google that allows you to build on-device AI applications. In this session, we will discuss what Coral is, what it can do, and the exciting updates and innovations in the Coral ecosystem.

🤔 What is Coral and what can it do?

Coral is a suite of hardware that integrates with Edge TPU, a tiny chip capable of performing machine learning inferencing at 4 trillion operations per Second. It allows developers to run machine learning models right on the device, without the need for cloud connectivity. Coral provides developers with an easy-to-use yet powerful tool to design innovative AI solutions.

💡 Examples of Coral applications

Object detection

Coral can perform object detection in real-time with impressive accuracy. Machine learning models running on Coral hardware can identify a large number of automobiles on a highway, as demonstrated by a live video feed.

Object classification

In addition to object detection, Coral can also perform object classification. It can identify different objects such as cars, buses, traffic lights, and pedestrians with high precision and speed.

Body part segmentation

Coral enables machine learning models to perform body part segmentation. By anonymizing people's images, Coral can track movements and gestures without compromising privacy.

Gesture recognition

With Coral, developers can identify people's gestures right on the device using machine learning models. This opens up possibilities for developing applications that can automatically respond based on customer's body movements.

Human figure detection

Coral can detect human figures in images and videos, enabling applications in augmented reality, animation, Fitness Coaching, and more. It can accurately determine which part of the human body appears in the image.

Facial detection

Using the MediaPipe model, Coral can perform real-time facial detection while preserving privacy. It can analyze facial expressions without identifying the person, enabling various applications that rely on facial Shape recognition.

🔥 Innovations using Coral

Coral has inspired a Wave of innovation in AI at the edge. Many companies, ranging from startups to large enterprises, have integrated Coral into their products and services. Here are a few examples:

Care.ai's self-aware healthcare environments

Care.ai specializes in building self-aware environments for healthcare facilities. They use Coral to develop AI sensors and cameras that monitor patient movements. By applying machine learning models, they can detect certain movements such as a patient falling and trigger alerts to medical staff, improving care while maintaining patient privacy.

Olea's smart water meter sensor

Olea has developed a smarter sensor for water meters using Coral's USB accelerated module. This solution allows machine learning models to analyze water usage in real-time at remote installation locations, providing independent checks for water meter accuracy without the need for cloud connectivity.

Hamburgini's AI cooking system

Hamburgini, a fast-growing restaurant chain in Saudi Arabia, has implemented an innovative AI cooking system using Coral. By tracking cooking time and temperature with machine learning models, they have significantly improved the quality of food served at their restaurants. Coral enables precise cooking and enhances the overall dining experience.

🆕 New products in Coral

Coral has introduced several new hardware products to expand its capabilities. These include:

M.2 modules

Coral's M.2 modules offer flexibility for developers by allowing them to easily add AI capabilities to existing systems. The modules come equipped with Edge TPU, enabling machine learning acceleration right on the device.

Coral Accelerator Module

The Coral Accelerator Module is a low-cost chip module that offers the same machine learning performance as other Coral products. It provides design flexibility for customers who want to integrate Edge TPU into their own host boards.

Coral Dev Board Mini

The Coral Dev Board Mini is a full-featured single-board computer with a smaller form factor and lower power consumption. It features onboard networking options and a Coral Accelerator Module, making it suitable for both prototyping and embedded deployment.

🖥 Updates in Coral software

In addition to hardware advancements, Coral has made significant updates to its software offerings. The machine learning model library has expanded to include 25 pre-compiled and ready-to-use models. The Edge TPU runtime has also been updated to version 13, now supporting Mendel Linux, MacOS, and Windows. The Coral platform provides developers with powerful tools and libraries to facilitate AI application development.

🤝 Coral partnership program

Coral has recently launched the Coral partnership program to foster collaboration and expand the on-device AI ecosystem. Product companies and service providers can join the program to receive support from Coral and promote their AI solutions. This partnership program aims to drive further innovation and adoption of Coral in various industries.

⭐ Introducing the new Coral Dev Board Micro

Exciting news! Coral is introducing the Coral Dev Board Micro, a microcontroller-class single-board computer with a tiny form factor. It offers the same level of machine learning acceleration performance as other Coral products but consumes very little power. The Dev Board Micro uses the FreeRTOS real-time operating system and supports TensorFlow Lite Micro machine learning models. With its small size and battery-powered capabilities, developers now have even more opportunities to create AI solutions at the edge.

👋 Conclusion and call to action

We hope this session has given you a comprehensive understanding of Coral and its capabilities. The Edge AI landscape is full of opportunities, and Coral empowers developers to create innovative AI applications that run directly on devices. Take the next step and embark on your Coral journey today. Join the Coral community, explore its resources, and start building the next generation of AI-powered solutions.

🔗 Resources for learning and using Coral

• Coral Documentation
• Coral Community
• Contact Coral Team

Highlights

• Coral is a platform for building on-device AI applications.
• It integrates Edge TPU, a tiny chip capable of on-device machine learning inferencing.
• Coral enables object detection, object classification, body part segmentation, gesture recognition, human figure detection, and facial detection.
• Innovations using Coral include self-aware healthcare environments, smart water meter sensors, and AI cooking systems.
• New products in Coral include M.2 modules, Coral Accelerator Module, and Coral Dev Board Mini.
• Updates in Coral software include an expanded model library and support for multiple operating systems.
• Coral partnership program encourages collaboration and promotes AI solutions using Coral.
• Introducing the Coral Dev Board Micro, a small form factor microcontroller-class single-board computer.
• Join the Coral community, explore resources, and start your Coral journey today.

Frequently Asked Questions

Q: What is Coral? A: Coral is a platform developed by Google for building on-device AI applications. It integrates Edge TPU, a powerful chip capable of performing machine learning inferencing right on the device.

Q: What are some applications of Coral? A: Coral can be used for object detection, object classification, body part segmentation, gesture recognition, human figure detection, and facial detection. It has been successfully implemented in healthcare, water usage analysis, and the food service industry, among others.

Q: What are the new products in Coral? A: Coral has introduced M.2 modules for easy integration into existing systems, Coral Accelerator Module for design flexibility, and Coral Dev Board Mini, a smaller and lighter single-board computer.

Q: How can I get started with Coral? A: You can refer to the Coral documentation and join the Coral community for resources, support, and collaboration opportunities. Start exploring the possibilities of AI at the edge with Coral.

Q: What is the Coral Dev Board Micro? A: The Coral Dev Board Micro is a microcontroller-class single-board computer with a tiny form factor. It offers machine learning acceleration while consuming very little power, making it ideal for battery-powered applications.

Q: How can I join the Coral partnership program? A: Both product companies and service providers can join the Coral partnership program. Contact the Coral team for more information and to explore partnership opportunities.