IEEE SPS Computer Vision and Machine Learning Tutorial

by AMD engineers from computer vision and machine learning group

7pm – 8:30pm, Tuesday, Dec. 15

Event organized and sponsored by

IEEE Signal Processing Society Santa Clara Valley chapter

IEEE Signal Processing Soceity

Schedule:

7:00 – 7:05pm Announcements

7:05 – 8:30pm Tutorial

Location: Online.

Please register on this page to get the meeting info sent on the day of the event.

Credit for SPS members who buy a Raspberry Pi for this tutorial:

This event is subsidized by IEEE Signal Processing Society (SPS). We will offer credits for SPS members who purchase a Raspberry Pi for this tutorial.

To be eligible for the credit, please register by Friday, Dec. 4th, and provide your IEEE membership number during registration. We will inform you of the exact credit amount. The proof of purchase is needed to receive the credit.

Join SPS today

Abstract

Part 1 – Machine Learning Tutorial

Data is present everywhere. It is up to us on how we use this data and derive new information out of it. Over the past few years Machine Learning and Artificial intelligence play a key role in our day to day activities. The computational capabilities of hardware also got innovated into many ways to get the possible optimal performance to train modern machine learning models. To apply machine learning, we have a lot of frameworks that support it. This part of the tutorial focuses on one specific Deep Learning Framework – Pytorch. Pytorch has got a tremendous popularity mainly because of its usability, and ready to deploy in production environments. In this tutorial, you would be learning some recipes to build various pipelines to be able to train a model from scratch and finally be able to convert Pytorch models to a popular exchange format like ONNX. Also, we will learn how to perform inference in Pytorch .

Hardware requirements – X86 Laptop/Desktop

Part 2 – Neural Net Exchange Formats

In Part 2 we will look into popular neural net exchange formats and how to leverage these formats in your neural net deployment stage

In our tutorial, we obtain an MNIST model in ONNX (Open Neural Network Exchange format; developers-Facebook, Microsoft) from part 1 (ML training). We convert this model to NNEF (Neural Net Exchange format).

Hardware requirements – X86 Laptop/Desktop

Part 3 – OpenVX

OpenVX™ is an open, royalty-free standard for cross platform acceleration of computer vision applications. OpenVX enables performance and power-optimized computer vision processing, especially important in embedded and real-time use cases such as face, body and gesture tracking, smart video surveillance, advanced driver assistance systems (ADAS), object and scene reconstruction, augmented reality, visual inspection, robotics and more

In this part we will learn how to create a simple application using OpenVX on Raspberry Pi and what are the advantages of using an open source, royalty free API in the deployment process.

Hardware requirements – 1 or 2

1. X86 Laptop/Desktop

2. Raspberry Pi 4 Model B Rev 1.2 / Raspberry Pi 3 Model B Rev 1.2

Raspberry Pi – Amazon Options

1. Raspberry Pi 4 official Kit – https://tinyurl.com/y2udhcxe
2. Raspberry Pi 4 – https://tinyurl.com/y5capenv
3. Raspberry Pi 3 – https://tinyurl.com/yykxjaas