Hi He,

using tf.depth_to_space is an alternative approach to increasing the height and width of a tensor for decoders. http://www.inference.vc/holiday-special-deriving-the-subpixel-cnn-from-first-principles/

It is sometimes referred to as SubPixel method, and it reduces artifacts that would otherwise come from strided deconvolution.

I used tf.space_to_depth instead of strided convolutions in the encoder in order to ensure symmetry in the network.

Hope that explains things!

PhD. Interests include Deep (Reinforcement) Learning, Computational Neuroscience, and Phenomenology.

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