Twitter | Pretraživanje | |
Pretraživanje Osvježi
Ming-Yu Liu 19. lip
We have released our under the BSD license (It was under CC non-commercial license.) Please feel free to use it.
Reply Retweet Označi sa "sviđa mi se"
William Ngan 21. sij
Recently I put together a small dataset of Bentley's classic snowflake photos, and trained a model. Here I generated some shapes with Pts.js and ran them through the model. Initial results -- (I probably need to train it some more but it's so expensive! 😭)
Reply Retweet Označi sa "sviđa mi se"
Mario Klingemann 21. sij 2019.
I've created an experimental GAN architecture I call or "Recursive-Residual GAN" and I am pretty astonished that: - it works at all - how well it works across a pretty wide range of scales. - it is just 15% the size of a comparable model
Reply Retweet Označi sa "sviđa mi se"
Ming-Yu Liu 15. ožu
Glad to see that our research works enable people to "generate realistic dance videos of NBA players for in-game entertainment." ,
Reply Retweet Označi sa "sviđa mi se"
samim 1. pro 2017.
The work by & is a hint at the future of design tools: Code coming soon:
Reply Retweet Označi sa "sviđa mi se"
Christian Mio Loclair 21. sij 2019.
Research at the lab | turning video games into interactive mode to question the value of designed pixels in future graphical productions | This is controlled with a body
Reply Retweet Označi sa "sviđa mi se"
Nono Martínez Alonso · Nono.MA ★ 16. lis 2018.
Odgovor korisniku/ci @nvidia
by "Synthesizing and manipulating 2048x1024 images with conditional GANs"
Reply Retweet Označi sa "sviđa mi se"
William Ngan 25. sij
1. Draw shapes in , by hand 2. Run with , fingers crossed 🤞
Reply Retweet Označi sa "sviđa mi se"
Mario Klingemann 31. pro
Odgovor korisniku/ci @FabLabMuc @BL_Labs i 4 ostali
By the end of 2017 my efforts to improve resolution were obliterated by two major breakthroughs: in short succession first showed their highly realistic celebrities made with and followed up with shortly after.
Reply Retweet Označi sa "sviđa mi se"
William Ngan 19. pro
Some generated snow crystals from GAN, based on Wilson Bentley's classic photos. Training is still in progress, slowly slowly. More to come soon!
Reply Retweet Označi sa "sviđa mi se"
A m m a r Ul H a s s a n 21. ruj
I personally think if this image is converted into segment ( how pix2pixhd works ) then i m sure var won’t make these types of errors. We need image to outline based concept here.
Reply Retweet Označi sa "sviđa mi se"
Mario Klingemann 8. srp
Odgovor korisniku/ci @quasimondo
Both models are shallow ResNets derived from . I first tried UNets, but there the models learned to cheat very quickly and just abused the first skip connection to pass the information almost uncompressed.
Reply Retweet Označi sa "sviđa mi se"
MEMO M̸e̵h̶m̸e̸t̶ ̴S̴e̷l̴i̷m̸ ̵A̸k̶t̴e̷n̶ 20. ožu
Thx 🙏 :). (To complete the loop, this 👇 is based on , coauthored by , also coauthor on . & other coauthors are from ).
Reply Retweet Označi sa "sviđa mi se"
Kyle Steinfeld 1. ožu
Some more developments in a series that extends work exhibited at 1/many
Reply Retweet Označi sa "sviđa mi se"
Kyle Steinfeld 5. velj
Some more initial results from a series that extends work exhibited at . Using streetview data, we trained a model to transform depthmap images to photographic images.
Reply Retweet Označi sa "sviđa mi se"
Mario Klingemann 21. sij 2019.
Odgovor korisniku/ci @quasimondo
The principle is pretty simple: in a classic residual architecture you chain several residual blocks behind each other (in the default is 9 blocks), what I do in is to use a single block, but loop 9 times over it, feeding its output back into its input.
Reply Retweet Označi sa "sviđa mi se"
JC Testud 7. stu 2018.
Reply Retweet Označi sa "sviđa mi se"
Mario Klingemann 27. lis 2018.
Odgovor korisniku/ci @ianholing
Thanks! There are 5 different GANs involved which employ my own architecture that owes a lot to and .
Reply Retweet Označi sa "sviđa mi se"
hans 27. kol 2018.
Hopping on the frame prediction train. Here's a feedback loop between 2 models, one trained to predict the next frame in a video, the other trained to predict the original frame from a version processed by the other model between 1 and 10 times.
Reply Retweet Označi sa "sviđa mi se"
Mario Klingemann 23. kol 2018.
Odgovor korisniku/ci @pretendsmarts
These are some snapshots from the training of my custom version of using on the Costică Acsinte archive.
Reply Retweet Označi sa "sviđa mi se"