From the course: Evaluating and Debugging Generative AI

Unlock the full course today

Join today to access over 23,400 courses taught by industry experts.

Challenge: Evaluate image quality

Challenge: Evaluate image quality

From the course: Evaluating and Debugging Generative AI

Start my 1-month free trial Buy for my team

Challenge: Evaluate image quality

“

Welcome to your first hands-on challenge. In this challenge, you'll use the FID score to evaluate the quality of images generated by a GAN. TensorFlow, an end-to-end platform that specializes in deep learning and makes it easy for you to build and deploy models, is our tool of choice. First, you'll need image data stored in two directories, one with real images and another with generated images. You can find sample data in the fake_images and real images directories. You'll use the tensorflow_gan library to calculate the FID score. Make sure to install it. Then you'll load the images from the images directory and pre-process them. Images are preprocessed to match the input requirements of the Inception V3 model, including resizing to 299 by 299 pixels, and normalization. The development environment you'll use during this course is fully online, so there's no need for you to install anything locally. Throughout this course, we'll use GitHub Codespaces, a cloud-hosted development…

Contents