Increasing Image Size

Hello,
What does increasing the image size do to the learning algorithm? And does increasing the image size make a copy of the original image and feeds that copy to the learning algorithm?

It depends a little bit If I understand it correctly. I would say that a different size does not do too much for the network per se, BUT e.g. if you train the same network on two different sizes of the same pictures the network might pick up something new in the newly sized pictures and it prevents overfitting. That what jeremy did in his dog breed notebook e.g. if you want to see an example with learn.set_data()

1 Like