ILSVRC 2012, commonly known as ImageNet, is a large image dataset for image classification. It contains 1000 classes, 1.28 million training images, and 50 thousand validation images. You can find more information about the dataset at or at

Style transfer using
Style transfer using
style transfer using

Why take a subset of ImageNet?

While the dataset is invaluable to the research community, its size prevents common researchers from conducting experiments on it. It requires more than 150GB of storage, and training a resnet50 on it will take around 215 hours using a T4 GPU on Google Colab, not to mention that Colab limits each session to 12 hours.

Taking a subset of ImageNet…

In 2014, Goodfellow et al. presented the Generative Adversarial Network (GAN) that generates images similar to the ones in the image dataset without the need for training labels. Ever since, the research community embraced GANs and new papers on GANs are released all the time. I am going to show you the basic structure of a GAN, and how you can build one to generate MNIST digits. The accompanying Colab Notebook is available here.

The GAN consists of a generator network (G) and a discriminator network (D), each of which is a neural network with potential convolutional layers. The discriminator…

Roland Gao

I am a UofT student interested in Computer Vision and Deep Learning.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store