Yahoo's model could be fine-tuned: http://caffe.berkeleyvision.org/gathered/examples/finetune_flickr_style.html http://caffe.berkeleyvision.org/gathered/example

Yahoo's model could be fine-tuned: http://caffe.berkeleyvision.org/gathered/examples/finetune_flickr_style.html http://caffe.berkeleyvision.org/gathered/examples/finetune_f... Yeah, it's definitely not a nice task but what's stopping someone (well, besides potential legal issues) from using these commercial APIs to create datasets programatically and training a cloned model from that? I'm curious what the profit margins are on these APIs because I think they are way overpriced.