Image Classification using SuperGradients and RegNet
What you will learn
Define how to design a search space for network architectures
Describe the AnyNet architecture and how it informs the RegNet design space
Describe the analysis and findings of the RegNet paper
Perform image classification using RegNet
Description
In this course, you are going to learn about RegNet architecture.
You’ll learn what it is, why it’s important, and the novelty that it introduced. With all of the free courses that I put on Udemy, I’m focused more on giving you the intuition and the kind of reasoning into the importance of this paper. If you’re looking for mathematical details, I’ll link to the paper and additional resources that you can go through. The thing about these free courses is that they’re limited to no more than an hour, which means I can only fit so much into this course.
You will learn a lot and you’ll walk away with a template for a project that you can use, cuz you will see RedNet inaction using the open-source training library called SuperGradients. It’ll provide a great foundation for you to just plug and play your own data sets and do a cool image classification project.
Here’s what the agenda is like for this course: Start off by talking about the motivation for RegNet: what it is, why is it that we’re embarking on this journey. And then from there you’ll learn about design spaces and then we’ll get into two specific design spaces, the AnyNet design space and the RegNet design space. I’ll link to resources that will give you a deeper understanding of these as well.
The course will wrap up with an overview of the analysis and findings that the researchers discovered in their experiments.
And then finally, we’ll see RegNet in action using the SuperGradients training library to perform image classification tasks.
Content