zechenghe
zechenghe
Very nice work! However, I was wondering how is E_{\tau} in the equation reflected in the implementation? It seems in the code only one (unshifted) image is used?
I was wondering if the deconv filters are trainable, or they are just transpose of the (pretrained) conv filters? If the latter, are we assuming the filters are unitary?
I was wondering if any experiments using pre-trained encoder as input_hint_block, for example directly use vae auto-encoder, rather than train this part from scratch? It would be beneficial for limited...
Why h_tv and w_tv are quadratic sum, not abs sum? Any difference in practice?