Getting Edgy with Machine Learning

How will you bring Machine Learning (ML) to edge devices to creatively solve challenges in IoT at the Edge? This is your chance to get edgy and show off your amazing new Machine Learning model and project for a chance to win this design contest.

We are looking for projects that demonstrate how ML can be used to solve problems in the fields of smart home, industrial, safety, or environmental. In this contest, we invite you to bring ML/AI to edge devices with the PSOC™ 6 AI Evaluation Kit and DEEPCRAFT™ Studio (formerly Imagimob Studio), our ML development platform.

We are asking contest participants to design their own edge AI model by:

1. Collecting sensor data using the PSOC™ 6 AI Kit onboard sensors (Note: We have removed external sensor data collection from the scope of the contest)

2. Building and training a new model inside DEEPCRAFT™ Studio (formerly Imagimob Studio) or leveraging one of the evaluation DEEPCRAFT™ Ready Models in an embedded application.

3. Showcasing how your model interacts with the physical world by building a project using ModusToolbox™ that deploys onto the PSOC™ 6 AI Kit.

4. Documenting your solution as a Hackster project and submit to the contest

Here are some application ideas to help you get started designing ML at the edge, but the sky's the limit.

We are curious to see how you will bring ML to edge devices! How edgy can you be?

The Hardware

Infineon’s PSOC™ 6 Artificial Intelligence Evaluation Kit (CY8CKIT-062S2-AI) is your perfect platform for easy prototyping and collecting real-life data to build compelling ML products fast. Being only the size of a cracker, the hardware platform focused on Edge AI and enables customers to evaluate Infineon’s Machine Learning platform DEEPCRAFT Studio, as well as ready to deploy ML models, and other software products.

Data collection is enabled via PSOC™ 6 MCU, as well as Radar (BGT60TR13C), high-performance digital MEMS microphone (IM72D128), Barometric Air Pressure Sensor (DPS368) & IMU sensors (Bosch BMI-270, BMM-350). Connectivity is enabled with WiFi/BT BLE Combo (Murata LBEE5KL1YN). Participants are encouraged to expand the hardware according to their use case, e.g. using additional sensors or data sources.

Opportunities exist for free hardware to a select number of applicants. (application has closed) or purchase here.

The Software

With ModusToolbox™ and DEEPCRAFT™ Studio (formerly Imagimob Studio) you can go through the full Machine Learning to Embedded Software flow from data collection and training to reference deployment.

DEEPCRAFT™ Studio (formerly Imagimob Studio)

DEEPCRAFT™ Studio is an end-to-end platform for developing AI / ML models for the edge. The platform is designed to support users in creating robust and high quality models that are ready for deployment in commercial products.

DEEPCRAFT™ Studio gives users the tools and flexibility required to customize their models for their specific needs. It is the fastest way for experts and non-experts alike to build production grade Machine Learning (ML) models for a wide range of use cases.

NOTE: DEEPCRAFT™ Studio is only available for Windows OS

DEEPCRAFT™ Ready Models

DEEPCRAFT™ Ready Models are the fastest way for Infineon's customers to take AI features into production. These fully tested, market-ready models are designed to give companies a way to add AI into products without developing machine learning models on their own.

DEEPCRAFT™ Ready Models are optimized to work with Infineon's sensors and MCUs.

DEEPCRAFT™ Ready Models use various sensors, and new models are always in development. There are currently six Ready Models available for the PSOC 6 AI Kit:

• DEEPCRAFT™ Ready Model for Baby Cry Detection
• DEEPCRAFT™ Ready Model for Cough Detection
• DEEPCRAFT™ Ready Model for Factory Alarm & Siren Detection
• DEEPCRAFT™ Ready Model for Gesture Detection
• DEEPCRAFT™ Ready Model for Siren Detection
• DEEPCRAFT™ Ready Model for Snore Detection

A helpful tutorial project "Using DEEPCRAFT Ready Models in ModusToolbox" is posted for your use as you develop your project.

ModusToolbox™ Software

Provided as a collection of development tools, libraries, and embedded runtime assets, ModusToolbox™ Software is architected to provide a flexible and comprehensive development experience.

Run-Time Software comprised of middleware, device drivers, and code examples is provided via an extensive collection of GitHub-hosted repositories. Explore the available run-time software resources cataloged within the ModusToolbox™ Software repository overview.

ModusToolbox development tools enable the creation of new embedded applications, managing software components, configuring device peripherals and middleware, and embedded development tools for compiling, programming, and debugging. These development tools interface directly with available run-time software repositories, providing easy access to the latest development resources.

Getting Started

The kit is pre-programmed with code examples to demonstrate sensor data collection from the USB port and store the data in the DEEPCRAFT™ Studio, ready to go for labelling and machine learning model creation. To get started, simply:

1. Connect to your PSOC™ 6 AI Kit and follow our Getting Started Guide

2. Download DEEPCRAFT™ Studio (formerly Imagimob Studio) (only available for Windows)

3. Download ModusToolbox™ (recommend v3.4 or newer)

4. Check out additional code examples (sound recognition/movement detection) featuring Imagimob ML models and cloud connectivity via Avnet IoT Connect.

5. Follow the steps found in these helpful tutorials:

• Updating the PSOC 6 AI Eval Kit Streaming Protocol Firmware
• Using DEEPCRAFT Ready Models in ModusToolbox

Helpful Webinars

Rapidly create AI/ML edge solutions using Infineon’s PSOC™ 6 AI Kit

Learn how to connect the contest hardware lnfineon's PSOC™ 6 Artificial Intelligence Evaluation Kit to the cloud using Avnet loTConnect on Azure and AWS. Presented by Sarah Hemmer.

WATCH HERE

Getting Started with ML Model Creation using the PSOC 6 AI Kit

How to get started with the devkit and valuable tips on optimizing development workflows. Presented by Clark Jarvis.

WATCH HERE

VIEW PRESENTATION SLIDES HERE

Prizes

Best Overall

Best Overall will be awarded to the project that demonstrates novel, well-engineered, and highly accurate ML project that solves a complex, real-world problem.

1 winner

$2500 USD e-GiftCard

Best ML Project - Radar

Best ML Project - Radar will be awarded to the project demonstrates exceptional performance in detecting and classifying targets, such as objects or people, with high accuracy and speed.

1 winner

$1000 USD e-GiftCard

Best ML Project - Motion

Best ML Project - Motion will be awarded to the project that has exceptional
performance in detecting motion (using the accelerometer, magnetometer, and gyroscope
sensor), such as gestures, human activity, or orientation measurements of the board.

1 winner

$1000 USD e-GiftCard

Best ML Project - Microphone

Best ML Project - Microphone will be awarded to the project utilizing the on-board microphones to detect voice commands, alerts, or other audible events.

1 winner

$1000USD e-GiftCard

Best Use of DEEPCRAFT™ Ready-Models

4 awards will be given to the best projects that leverage the DEEPCRAFT™ Ready-Models in a
unique way.

4 winners

$500 USD e-GiftCard

Judges

Clark Jarvis

Clark is a senior software technical marketer at Infineon. Building on nearly two decades of product experience for embedded software and development tools, Clark focuses on creating efficient and intuitive design experiences to help embedded engineers ultimately master and speed the development of IoT and industrial applications.

Sarah Hemmer

Sarah is a senior software product manager at Infineon, always looking to take exciting new products and ideas to market. Coming from a background in behavioral economics and data analytics, she will focus on bringing the customer perspective to this contest.

Nick Sharp

Nicholas is a Staff Applications Engineer at Infineon in the MCU division. Currently he is working on integrating machine learning into embedded systems. Nicholas holds a Computer Engineering degree from Seattle Pacific University.

Sam Al-Attiyah

Sam has been working at Imagimob for more than 7 years. He has a Master in Electrical Engineering having studied both in Australia and in Sweden. Over this time he has had experience working with the end-to-end machine learning applications on the edge and led multiple projects to production. Sam is now heading the Customer Success team at Imagimob which focuses on ensuring customers get to production with their machine learning applications. Sam is also the product manager for the DEEPCRAFT Ready Models which are fully trained models ready deployment and commercialization.

Resources

About us

Infineon is a world leader in semiconductor solutions that make life easier, safer, and greener. Our solutions for efficient energy management, smart mobility, and secure, seamless communications link the real and the digital world.