Yingtao Tian
3,847 posts
Yingtao Tian
@alanyttian
↑ profile picture is dreamed by Anime GAN /
cooking computational and ML sauce at Sakana AI /
before: google {brain,deepmind} ← stony brook u ← fudan u
- ARC Ukiyo-e dataset faces (codh.rois.ac.jp/ukiyo-e/face-d…) can be translated from selfies (github.com/znxlwm/UGATIT-…), if we train a GANsNRoses model (github.com/mchong6/GANsNR…) on them......
00:00 - On the wall outside NeurIPS Workshop on Machine Learning for Creativity and Design 3.0
- Joining Sakana AI today.
- Generating Virtual YouTubers face, fully controlled by pose parameters (or even pose extracted from existing videos), from only a single image rather than 3D model. Very interesting work by Pramook Khungurn. pkhungurn.github.io/talking-head-a…
- Modern Evolution Strategies for Creativity: Fitting Concrete Images and Abstract Concepts It's a fun work w/ @hardmaru Pdf arxiv.org/abs/2109.08857 Web es-clip.github.io Code github.com/google/brain-t…
GIF - This summer I will join Google Brain *Tokyo* for a research position after graduation. I am super excited, seeing many opportunities in research and beyond! Also it seems to be time to consider adding a "views are my own" disclaimer 🤣
- After defending my thesis --- I'm (officially) a PhD Now 😀
- youtube.com/watch?v=PUkQbG… 10 Minutes of Imaginary Japanese Anime Face, dreamed by (an 256x256 version) of our #MakeGirlsMoe GAN model that is accepted in #NIPS2017 Workshop for Machine Learning for Creativity and Design!
00:00 - This is *amazingly* great in terms of quality!
- Our work "Ukiyo-e Analysis and Creativity with Attribute and Geometry Annotation" has been accepted #iccc21! Ukiyo-e paintings with labelled attributes and automatically extracted face landmarks, allowing quantitative analysis and fun ML experiments. arxiv.org/abs/2106.02267
- Image transfer quality that i've being dreaming of.京都旅行中に写真を撮った。
- Our work "Kaokore: The Pre-modern Japanese Art Facial Expression Dataset" has been accepted @iccc_conf '20 ! Looking forward to presenting it, either physically on Portugal, or more likely virtually.KaoKore: A Pre-modern Japanese Art Facial Expression Dataset Labelled faces extracted from pre-modern Japanese artwork! Allows for fun ML experiments, like Intrinsic Style Transfer + Learning to Paint. Researchers are probably “commoners” in Edo-Period. arxiv.org/abs/2002.08595
