Deep Neural Analogies
Finally got some interesting results out of the code example Microsoft research released for their Visual Attribute Transfer through Deep Image Analogy paper.
Don't have time to go into details but I'll do a writeup on this later, for now I just wanted to share some results.
Style image
"miss universe" by Silvia Pelissero
Reverse style
Result with the old algorithm (using neural-style-tf) for the same inputs
Turns out to be really hard to get the output you're looking for, have to get the parameters and style/content image combo exactly right. Here's some of the more interesting failure cases:
Have a nice weekend!
wow, I wasn't quite getting how cool this is until I visited the github repo! geez, pretty amazing stuff.
Yeah the evolution speed of the neural style algorithms are mind-blowing
Interesting results! I also want to do some experiments using these techs.
Thanks for sharing this news : ) really excited to see the advancement of AI tech. Being one of the leaders in the tech industry, Microsoft has been doing a lot of researches in AI. The papers published by Microsoft Research are nice. There seems to be a lot of state-of-the-art applications of AI on art work. Would be nice to see more about this
Wow, I don't even have words to describe how amazing this looks!
Are these images generated by a program? Thank you for you efforts to provide us value :) You have a new follower. I will appriciate it if you follow me too :))
Thanks, yeah they are generated by an artificial intelligence, check paper linked if you're interested :)
So amazing! Great work
very nice art of work Followed waiting for more great stuff from you :)
I like it! :)
I can't say I understand the logic behind it. But the results look promising. I am not sure that I am the only human commenting that by the way. :))
Haha yeah many bots on here now. You seem real though :)
wow that's great! could you check out my account??