Hello trend,
This is Satya Mallick from LearnOpenCV.com.
In today's blog post we will learn about Variational Autoencoder in TensorFlow. Before Generative Adversarial Network (GAN) was invented, there were other neural network architectures for Generative Modeling. Today we will take you back in time and discuss one of the most popular pre-GAN eras Deep Generative Model known as Variational Autoencoder.
If you have never heard about Variational Autoencoders, this is just the right post for you because we start from the very beginning and explain the concept with code. We are going to build on Autoencoders we covered last week, discuss the important ideas that make variational autoencoders different from plain vanilla autoencoders, and finally have fun experimenting with code, and two different datasets - Fashion MNIST & Google Cartoon Dataset.
https://learnopencv.com/variational-autoencoder-in-tensorflow/
and experiment with the code at the following link
https://github.com/spmallick/learnopencv/tree/master/Variational-Autoencoder-TensorFlow
From Training to Deployment in One Big Step : Live Coding Tutorial
As part of OpenCV Weekly Webinar, we are bringing world class practitioners to discuss practical problems and solutions related to computer vision and artificial intelligence. Last week's episode was a discussion on solutions to common pitfalls for Video AI with guest Tyler Compton.
This week we will have a live coding session with Raymond Lo who is the OpenVINO Edge AI Software Evangelist at Intel. He will teach us how to train a model on Google Colab and then deploy it on an edge hardware (OpenCV AI Kit)
Topic : From Training to Deployment in One Big Step
Time : 9 AM Pacific Time, April 29, 2021
Hosts : Raymond Lo (Intel), Satya Mallick (CEO, OpenCV.org), and Phil Nelson (Content Manager, OpenCV.org)
Free registration : Zoom Registration Link
Resources : OpenVINO notebooks, image classification example, Collab notebook.
Hope to see you there!
How do we keep LearnOpenCV Free
Most of the posts we write on LearnOpenCV.com take anywhere between 40-80 hours of work. We do not run any ads on our site and never will. There are two ways we are able to support free content on the site.
Consulting Services
My consulting company, Big Vision, provides consulting services in computer vision, artificial intelligence, and machine learning. We work with well funded startups, medium and large sized companies to develop state of the art computer vision solutions for their use cases. If your company needs AI experts for solving an AI problem of medium to large size ($30k or more), please put in a good word for us. We can be reached at contact@bigvision.ai.
Official AI Courses by OpenCV.org
Our team is also responsible for creating and managing the Official AI Courses by OpenCV. These courses not only pay for our staff creating the courses, and free blog posts, they also help development of the OpenCV library.
Learning computer vision and deep learning is hard. Free blog content on the internet is great, but blog posts by their very nature, do not provide the structure, and depth needed for someone looking for mastery in the subject.
Even if you do find the right resources, they're often too advanced for beginners or don't teach practical applications that you can apply immediately in your projects.
We developed these courses to help fill this gap by teaching the basics of computer vision from scratch using simple English that anyone can understand. In our courses we first build a solid theoretical foundation before moving onto more advanced topics like machine learning and deep learning. We familiarize you with the right libraries, and show you how to build real-world projects!
You can buy the courses right away from our course site, or join the discount waitlist.
That is all for this week. Happy learning!
