Code for the paper: Foveated Convolutions: Improving Spatial Transformer Networks by Modelling the Retina