Semi-supervised StyleGAN: Disentanglement is through mutual information loss. Propose new metrics for measuring disentanglement in generator. Take-away: small amount of supervision is enough for disentanglement and high-res generation @NVIDIAAIpic.twitter.com/FDh26WbiQQ
-
-
Show this thread
-
Semi-supervised StyleGAN image-to-image: for semantic fine-grained image editing with strong generalization to unseen images.
@NVIDIAAI Featuring#DeepLearning leaders@drfeifei@geoffreyhinton@goodfellow_ian Jitendra Malik Antonio Torralba Yoshua Bengio (not in training data)pic.twitter.com/tkYRQ8onOxShow this thread -
We also create synthetic high-res datasets for benchmarking disentanglement learning using Falcor renderer.pic.twitter.com/5YHKoYAqrW
Show this thread -
We also used
@NVIDIAAI#Isaac Simulation environment to create synthetic high-res datasets for benchmarking disentanglement learning in semi-supervised settingpic.twitter.com/QGcS68h4ESShow this thread -
Slides for semi-supervised StyleGAN http://shorturl.at/crDMO
@wn8_nie@animesh_garg@shoubhikdn@killerooo@NVIDIAAI (link I posted previously doesn't work)Show this thread -
Project details on semi-supervised StyleGAN athttps://sites.google.com/nvidia.com/semi-stylegan …
Show this thread
End of conversation
New conversation -
-
-
Glad to have been part of this work!
-
Sorry
@killerooo Didn't locate your Twitter handle. Indeed, was great to collaborate with you at@NVIDIAAI
End of conversation
New conversation -
-
-
It looks like the URL to the slides is broken
-
Sorry, not sure why. This should workhttp://shorturl.at/crDMO
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.