TinyML on Arduino

With a mission to make machine learning (ML) simple enough for anyone to use, Massimo Banzi at Arduino will explain why you might want to squeeze ML on microcontrollers, including function, privacy, efficiency and cost.

He'll also cover how billions of microcontrollers combined with all sorts of sensors in all sorts of places will lead to innovative TinyML applications in the future.

Discover how ML can make microcontrollers accessible to developers who don’t have a background in embedded development.

IoT Embedded Vision Systems: Deploying Machine Learning Models from the Cloud to the edge

When developing an IoT embedded vision system for machine learning (ML) applications, find out how Basler, NVIDIA and AWS have bundled their competencies in the field of computer vision, ML and IoT to tackle key challenges and overcome complexity.

Discover how an optimised embedded vision architecture can be realised for the deployment of ML models from the AWS cloud to the edge device, enabling you to start prototyping smart, connected vision applications quickly.

Privacy-Preserving AI with Coral Edge TPUs

Coral, Google’s Edge TPU platform, will be used to show how moving your AI workload to the edge can help reduce latency, save power, eliminate/reduce bandwidth needs, and ensure privacy preservation.

Building Business Value with Intel AI and Industrial IoT at the Edge

In this introductory session, learn how you can create greater business value by leveraging Artificial Intelligence and IIoT by introducing you to terms, technologies, and uses of AI and IIoT.

By understanding the fundamentals, Intel helps you realise what benefits and market potential these two essential technologies can have in your business and the modern business ecosystem.

Making inferencing easy at the edge, without the power trip!

FPGAs are ideal for edge AI applications, especially inferencing in power-constrained compute environments. The reason: FPGAs can perform more giga operations per second (GOPS) with greater power efficiency than a central processing unit (CPU) or graphics processing unit (GPU). However, they require specialised hardware design skills.

Here we will explore how Microchip’s VectorBlox Accelerator SDK and Neural Network IP offer an easy way to create low-power FPGA smart embedded vision solutions without prior FPGA experience.

Navigating BittWare’s FPGA cards for acceleration

Now part of Molex, BittWare specialises in FPGA acceleration market where Intel and Xilinx programmable devices are now essential for accelerating compute-intensive or latency-critical workloads, such as machine learning inference or real-time data analytics.

In this presentation, we will help you navigate the range of BittWare FPGA acceleration cards available from Mouser. We’ll look at card options, intended industry segments and case studies.

NXP Solutions for Artificial Intelligence and Machine Learning

As AI methods gain maturity, we see interpretation and decision-making migrate from industrial PCs or datacenters into embedded devices. This change puts new demands on both the embedded processors and the tooling used to design and implement these systems.

In this demo, we examine why industrial applications benefit from embedded AI, and how NXP devices support the shift of intelligence to the edge.

AI for human body detection in industrial use cases

In this presentation, ST explains the need for Human Body Detection in various industrial applications. Common scenarios include detecting a person in danger or someone who has fallen, to detecting the presence of lone workers, or controlling safety around industrial equipment from workers clothing, and even visual wake-up-word recognition and face ID technology.

ST will demonstrate and show how the dataset is defined, how the neural network topology is chosen and built, and then how the hardware implementation is done on a microcontroller using STM32Cube.

Deep learning inference at the edge on TI processors

TI Processors are highly integrated, industrial-grade SoCs designed to run deep learning and inference at the edge—delivering the benefits of low power and low latency.

In this presentation, you will learn typical steps involved in deploying a deep learning model on any embedded processor. We will also take a closer look at the deep learning and inference development environment, plus look at the scalability of the solution across TI's portfolio. We will even explore the ease of usage with acceleration of using an open-source deep learning/inference engine, such as TFLite and Neo Run Time on TI processors.

AI Facial Recognition Demo

This project will demonstrate AI facial recognition using the Intel Neural Compute Stick 2 (NCS2) and the Raspberry Pi 3 Model B+. The Raspberry Pi will host a web camera, HDMI monitor, and interface with the NCS2 device. The NCS2 device will run a neural network to detect a human face, additionally estimating age, gender & emotion.

View project BOM »

AI facial recognition Demo

This project will demonstrate AI facial recognition using the Intel Neural Compute Stick 2 (NCS2) and the Raspberry Pi 3 Model B+. The Raspberry Pi will host a web camera, HDMI monitor, and interface with the NCS2 device. The NCS2 device will run a neural network to detect a human face, additionally estimating age, gender & emotion.

AI hand-written digit recognition with the Hello FPGA Kit

In this demo, we’ll explore how Microchip’s Hello FPGA kit can be used in the world of AI. We’ll show the kit built as an image recognition system running a Convolutional Neural Network (CNN) capable of recognizing hand-written digits. We’ll take a deeper dive into the CNN, explain how it works, and the limitations of this implementation.