Papers of the day   All papers

Progressive Face Super-Resolution via Attention to Facial Landmark

Comments

ΜΔDΞRΔS: #CyberpunkisNow A new AI/algorithm can accurately reconstructs faces from tiny 16×16 pixel input images. Top row are low resolution images, middle row are the AI's output, bottom row are the original photos. More Info- https://iforcedabot.com/photo-realistic-emojis-and-emotes-with-progressive-face-super-resolution/ https://arxiv.org/abs/1908.08239 https://t.co/gbHXNSYhcS

34 replies, 787 likes


roadrunner01: Progressive Face Super-Resolution via Attention to Facial Landmark pdf: https://arxiv.org/pdf/1908.08239.pdf abs: https://arxiv.org/abs/1908.08239 github: https://github.com/DeokyunKim/Progressive-Face-Super-Resolution https://t.co/xU5UsHTMJx

1 replies, 41 likes


Jonathan Fly 👾: I tried the new 'Progressive Face Super-Resolution via Attention to Facial Landmark' Funny detail not in official samples: a whole lot of Harry-Potter-Style scars. 3 out of the first 8. Paper: https://arxiv.org/abs/1908.08239 Code: https://github.com/DeokyunKim/Progressive-Face-Super-Resolution (note typo in eval py line 37) https://t.co/0mQMhrl2cW

3 replies, 32 likes


arxiv: Progressive Face Super-Resolution via Attention to Facial Landmark. http://arxiv.org/abs/1908.08239 https://t.co/YF677Z4QGU

0 replies, 17 likes


Supreme chancellor of antifa🏴🚩: Something to remember

1 replies, 17 likes


Paul Kedrosky: tl;dr: Don't go around pixellating those private images of yours anymore and thinking you've fooled anyone. /v @erikbryn Progressive Face Super-Resolution via Attention to Facial Landmark https://arxiv.org/abs/1908.08239 https://t.co/Xz7xPnr09Z

1 replies, 15 likes


Gabe: The steady march towards AI dystopia continues.

0 replies, 12 likes


Phreakops: fucking hell!!

1 replies, 9 likes


Christie Malry ☭ Ⓐ🏴‍☠️🏴🚩: fuk, mask up comrades

1 replies, 6 likes


Jannis Tenbrink (he/they): This is as scary as it‘s important. What that software needs is finetuning in how to interpret specific things, like shadows not being moustaches but overinterpreted artefacts by the pixelation. Then this would be close to flawless

1 replies, 4 likes


Sampson: "Progressive Face Super-Resolution via Attention to Facial Landmark," or generating high-resolution faces from 16x16 images. This approach is more accurate, yet uses only 1/3 the training time of other leading approaches. https://arxiv.org/pdf/1908.08239.pdf https://t.co/CLNhsra7Nw

0 replies, 4 likes


Philipp Häfelfinger: Ok, this is scary 😳

0 replies, 3 likes


Gerardo Montoya aka Rénròu Sōusuǒ: pixel life #sujetoshíbridos

0 replies, 2 likes


Amine Horse-man: Super-Resolution algorithms got so efficient they can now reconstruct blured and pixelized faces #AI #MachineLearning #DeepLearning

0 replies, 1 likes


raihan saputra: Say goodbye to privacy by blurring faces. The only good way is to not have any face in the frame. US military photos usually block faces with a black square; probably the only secure way. (Even then, needs good metadata cleaning to ensure security.)

1 replies, 1 likes


KΛPTEN™: be afraid.. be very afraid.. https://t.co/kq6fx1LHti

0 replies, 1 likes


The Kebetu Talk: Source : https://arxiv.org/abs/1908.08239

1 replies, 0 likes


Content

Found on Sep 02 2019 at https://arxiv.org/pdf/1908.08239.pdf

PDF content of a computer science paper: Progressive Face Super-Resolution via Attention to Facial Landmark