That’s a good idea!
Antony Harnist

Yes, in the article I provide snippet of code for the function flip_image(). A two-step process which is randomly flip the image on a vertical axis and then perturb the image. You can generate a lot of new training data on the fly without storing too many images on disk.

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.