Pink Floyd (AI Generated Album)
Альбом был создан с использованием https://openai.com/blog/jukebox/ и образцов Pink Floyd

https://youtu.be/IOJzybDCG6M

Pose Animator - инструментом веб-анимации, который оживляет иллюстрации SVG с помощью моделей восприятия человеком TF.js в реальном времени.
pose-animator-demo.firebaseapp.com

Github и документация по техническому проектированию:
github.com/yemount/pose-animator

Azulejo

https://twitter.com/cunicode/status/1259149413144027137?s=20

GANs on the library's digitised map collections

Model training as reconstruction, timelapse

https://twitter.com/erikswahn/status/1259005787311128576?s=20

Epipolar Transformers
pdf: https://arxiv.org/pdf/2005.04551.pdf
abs: https://arxiv.org/abs/2005.04551
github: https://github.com/yihui-he/epipolar-transformers

Machine Hallucinations : Urban Dreams

https://twitter.com/refikanadol

“Time is relative; its only worth depends upon what we do as it is passing.” ― Albert Einstein

RL agents get specific to tasks they are trained on. What if we remove the task itself during training?

Turns out, a self-supervised planning agent can both explore efficiently & achieve SOTA on test tasks w/ zero or few samples in DMControl from images!

https://ramanans1.github.io/plan2explore

FaceFilter: Audio-visual speech separation using still images

pdf: https://arxiv.org/pdf/2005.07074.pdf
abs: https://arxiv.org/abs/2005.07074
video: https://youtube.com/watch?v=ku9xoLh62E4

Single-Stage Semantic Segmentation from Image Labels

Use save model GAN in Unreal Engine 4

SketchRNN