Image Super-Resolution GANs
Using Python and Tensorflow 2.0, you can use Generative Adversarial Networks to improve or upsample images.
What you’ll learn
Image Super-Resolution GANs
- Create a generator that upsamples an image by 4 times in each dimension, so that it looks better.
- Create a discriminator that scores both realism and how well the image looks like the original.
- Keras layers can be changed so that they can accept images of any size without having to re-create the model.
- Google CoLab lets you train the models on a Cloud TPU through a TPU on the Cloud.
- Use the trained generator in a real-world application to make your own images bigger.
Requirements
- “High-Resolution Generative Adversarial Networks (GANs)” is the name of my class.
- Python is a language I know.
- Experience with convolutional neural networks.
- People who have some experience with TensorFlow 2.0 and Keras.
Description
We’ve all seen the TV show gimmick where the investigators can take a small piece of an image and make it look like it’s much bigger than it is. This is because today, Generative Adversarial Networks are making things that were once impossible possible.
Here, I’ll show you how to build on what you learned in my High-Resolution Generative Adversarial Networks course to achieve this impressive feat known as “Super-resolution.” If you train a Generator, not only will you be able to make an image four times bigger than it was before, but it will also be very easy for us to do.
Our convolutional neural networks will be built and trained with Python and Keras in the same way as in the first course. We’ll use Google CoLab again to connect to a free Cloud TPU to train our networks. You can finish the course in just a few days and not pay for any hardware!
To see if this sounds interesting, watch the free preview of the “Results!” lesson for a while. If you go, I have no doubt that you’ll be very happy.
Who this course is for:
- Python and TensorFlow 2.0 developers who want to make images look more realistic and clear when they get bigger.
