shoes dataset

[mxkai]

Can you tell me how to train on the edges2shoes dataset

[jbmaxwell]

I'm also curious about training for more general pix2pix applications, like the shoes and handbags examples on the compvis site—specifically, cases that don't use a human figure as the conditional input.

More specifically, I'm actually interested in distorting/reshaping specific images—so I would provide both a "sketch" and an image, but the sketch wouldn't be a list of joint coordinates.