8:30am — Building Neuromorphic Models with BrainChip’s Akida

Nikunj Kotecha and Chris Anastasi

Learn how to leverage BrainChip MetaTF to convert traditional CNNs to spiking neural networks (SNNs) with ease and apply to real world use cases.

• Ability level: Beginner to advanced ML engineers

• Prerequisites: Python, TensorFlow, and ML knowledge a plus

• Hardware: Software simulation provided

8:30am — Edge AI Connectivity Made Simple with Blues Wireless

TJ VanToll and Paige Niedringhaus
‍
If you’ve ever tried to run your ML applications outside of comfortable Wi-Fi range, you know how difficult IoT connectivity still is in 2022. The Blues Wireless Notecard solves this problem elegantly, by offering subscription-free cellular connectivity, on a secure system-on-module with a simple JSON API. The Notecard is the perfect companion to your ML applications, letting you focus on polishing your models, while allowing the Notecard to handle your connectivity needs.

In this workshop you’ll see how it all works. You’ll start by going through a hands-on tutorial to learn about the Notecard, and then see a real-world example of how to do ML inferencing on the edge while sending your results to a cloud dashboard. Plus, we might have an exciting new feature to demo for you live.

• Ability level: Intermediate

• Hardware: No physical hardware required, workshop will use a simulated Notecard. If you would like to follow along with real hardware, you will need:
  
  - Feather Starter Kit for Swan
  - Adafruit LIS3DH accelerometer
  - Qwiic cable

10:15am — Running AI Models on the Nvidia Jetson CPU and GPU, with balena and Edge Impulse

Marc Pous and Alan Boris
‍
In this workshop, balena's Marc Pous, Developer Advocate, and Alan Boris, Hardware Hacker in Residence, will showcase machine learning fleet management using the Nvidia Jetson Nano on the balenaCloud device management platform. The Jetson Nano will run AI models on both the CPU and the GPU, and we'll also show how to re-train and update the model across the fleet of devices. If you have a Jetson Nano, bring it to the virtual workshop to participate in the live-build of the fleet.

• Prerequisites: Knowledge of Docker is a plus

• Hardware: Nvidia Jetson, NanoCameraSD, CardLaptop adapter for SD card (if no internal port)

10:15am — Running Sensor Models on Syntiant's TinyML Board

Atul Gupta
‍
Learn how to run sensor models on Syntiant's TinyML Board.

• Prerequisites: Successful completion of go/stop keyword tutorial on the TinyML Board

• Hardware: TinyML Board required

12:00pm — Combining ModusToolbox and Edge Impulse to create and Deploy Acoustic Event Classifications on PSoC 6

Danny Watson and Nick Sharp
‍
The Infineon PSoC 6 is a highly capable Dual Core device that has been launched into the Edge Impulse Ecosystem. In this session we will provide an acoustic classification use case and go through the firmware compilation through ModusToolbox, on-device data collection, segmentation, labeling, training and deployment to do inferencing. By the end of the session you will have a model running on the PSoC 6 and prizes given to the most innovative and reliable deployments.

Ability Level: None

Hardware/Software: In advance of the workshop, please purchase and download:
• PSoC 6 Wi-Fi BT Prototyping Kit
‍• Install ModusToolbox

12:00pm — Train and Deploy your Own AI Model into a Vision AI Sensor

Lakshantha Dissanayake
‍
Artificial Intelligence camera (AI camera) is an enhanced camera powered by a built-in edge machine learning algorithm, smartly processing with computational photography to perform enhanced object detection in real-time. It has been widely used in smartphones for face recognition, edge devices for wildlife detection, and other edge intelligence applications. Featuring the Vision AI camera, Seeed Studio released the latest sensor prototype kit that includes the Grove - Vision AI module as the highlight to bring edge computing to IoT sensors. The module is a compact AI camera that supports simple ML model training and implementation thanks to the support from Edge Impulse. In this workshop, Seeed Studio will provide step-by-step guidance on how to train your own AI model for your specific application with Edge Impulse, and then deploy it easily to the Grove - Vision AI module to create you own Vison AI sensor.

Ability Level: Any skill level

Hardware/Software:
• SenseCAP K1100 - The Sensor Prototype Kit (Note: This kit is discounted and has free global shipping until September 30)

1:30pm — Hardware Accelerated ML on Alif Semiconductor's Ensemble Devices Takes Edge Processing From Watts to Milliwatts

Henrik Flodell

Alif Semiconductor recently introduced the Ensemble family of Microcontrollers and Fusion processors. These devices are the first general purpose controllers in the market to feature Arm’s Cortex-M55 MCU core, as well as the Ethos-U55 microNPU, which is design specifically for accelerating machine learning operations on deeply embedded devices.

In this session we will introduce the Ensemble family, and showcase the uplift it brings to Voice and Vision use cases in particular, in terms of raw performance and power savings.

2:30pm — The Fully Self-Contained Embedded ML Prototyping Solution

Robin M Saltnes
‍
The Nordic Thingy:53 is the first fully self-contained prototyping solution for embedded ML with complete wireless Edge Impulse integration through Bluetooth Low Energy. It takes the simplicity of getting started with embedded machine learning to a whole new level.

In this tech talk, Robin from Nordic Semiconductor will walk you through how easy it is to set up the Thingy:53 to communicate with the Edge Impulse cloud and use it both for uploading new training data and deploying ML models in the field. He will also give a closer look at the core components of the Thingy:53, including all of its built-in sensor hardware, and the key feature of the nRF5340 dual-core wireless SoC that enables both the ML capabilities and wireless connectivity.

3:00pm — Smart Sensors for Smarter Spaces

Chetan Joshi
‍‍
ML at Edge is and will be increasingly prevalent in ambient intelligence. In this talk, the attendees will hear from Panasonic about the fundamental shifts undergoing in the smart building industry. Using the example of Panasonic Grid-Eye infra-red matrix sensor, we will underline how simple, low-cost sensors can be improved using ML techniques, giving rise to deep insights into utilization of space.

3:30pm — Jump Start Your Edge AI Vision with TI

Reese Grimsley
‍‍
Machine Learning Inference has thoroughly penetrated embedded applications, especially with visual and spatial data like images and radar/lidar. Texas Instruments has a growingly scalable processor portfolio for vision inference at the edge. Arm-only execution on the Sitara AM62 and accelerated execution on the Jacinto TDA4VM provides a range from 0.5 to 8 TOPS of performance in a Linux or RTOS environment. With a large collection of pre-optimized AI models, no-cost and low-cost development tools, and a hardware-agnostic software programming environment, TI helps you bring your idea to realization on an embedded device in no time. With Edge Impulse, we are further democratizing edge AI application development to bring embedded inference closer to the community by simplifying AI model development for TI processors. Please join our Tech Talk to learn how TI and Edge Impulse can accelerate your vision for machine learning at the edge.

3:45pm — Accelerating Machine Learning at the Edge with Silicon Labs recently introduced MG24

Dan Kozin
‍‍
Silicon Labs recently announced our EFR32MG24, the leading wireless SoC for multi-protocol applications.  It also has an integrated AI/ML accelerator to offload machine learning operations for more optimized power applications. Edge Impulse’s Studio integrates this accelerator, enabling more complex use cases on this tiny edge device.  In this talk, we review applications enabled by the EFR32 AI/ML accelerator and also demonstrate object recognition using Edge Impulse’s FOMO block running simultaneously with wireless connectivity.

4:30pm — Extreme Edge Predictive Maintenance

Nathan Verrill and Mike Peacock

Collect multi-sensor data at scale, train once and deploy everywhere using Kafka, the Edge Impulse Ingestion API, and Custom Deployment Blocks to run EI models in Kafka Streams, in the cloud, and on devices, online and off, all from one Edge Impulse project.

5:30pm — The AWS Marketplace

Ankur Srivastava

The AWS Marketplace makes it easy for developers to discover and procure digital solutions so you can you build your next-best product.

9:00am — Opening, Daily Agenda

David Tischler, Development Program Manager, Edge Impulse

9:15am — Edge Impulse Announcements Deep Dive Session

Jan Jongboom and Arun Rajasekaran, Edge Impulse

10:00am — Recyclable Materials Sorter with NVIDIA Jetson Nano

Constantin Craciun, Zalmotek

10:30am — Adding AI Capabilities to Nanosaur

Raffaello Bonghi, Nvidia,
co-founder Pizza Robotics
Giovanni Di Dio Bruno, Università degli Studi di Napoli Federico II,
co-founder Pizza Robotics

11:00am — Seeed Studio Community Discussion

Eric Pan, founder and CEO,
Seeed Studio

11:30am — Tiny Robotics: Using MicroROS with Edge Impulse

Avi Brown, Edge Impulse Expert

12:00pm — What’s new with Azure IoT to get you building solutions faster

Jim Bennett, Regional Cloud Advocate, Microsoft

12:30pm — Measure Analog Gauges with Computer Vision Using a Sony Spresense

Mithun Das, Edge Impulse Expert

1:00pm — Machine Learning in the Wild: Tales from the Real World

Angus Thomson, CEO and co-founder, Canairy AI
Walt Jacob, CTO and co-founder, Canairy AI

1:30pm — Machine Learning on the Edge with TensorFlow Lite

Paul Ruiz, Developer Relations Engineer, Machine Learning, Google

2:00pm — Community Fireside Chat with Hackster

David Tischler, Development Program Manager, Edge Impulse
Jessica Tangeman, CEO, Hackster
Alex Glow, Lead Hardware Nerd, Hackster
Jinger Zeng, Contest Manager, Hackster