Zobrazeno 1 - 5
of 5
pro vyhledávání: '"Oran Gafni"'
Publikováno v:
Lecture Notes in Computer Science ISBN: 9783031197833
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_________::b2d843ed2c86db806dfeef66e02c16c8
https://doi.org/10.1007/978-3-031-19784-0_6
https://doi.org/10.1007/978-3-031-19784-0_6
Autor:
Daniel Haziza, Thibault Peyronel, Peizhao Zhang, Maxime Oquab, Oran Gafni, Tao Xu, Patrick Labatut, Camille Couprie, Yana Hasson, Onur Celebi, Bobo Bose-Kolanu, Pierre Stock
Publikováno v:
CVPR Workshops
To unlock video chat for hundreds of millions of people hindered by poor connectivity or unaffordable data costs, we propose to authentically reconstruct faces on the receiver's device using facial landmarks extracted at the sender's side and transmi
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::57e17ee435fcc8a5dab3a35202f36fac
http://arxiv.org/abs/2012.00328
http://arxiv.org/abs/2012.00328
Autor:
Oran Gafni, Lior Wolf
Publikováno v:
CVPR
We present a novel method for inserting objects, specifically humans, into existing images, such that they blend in a photorealistic manner, while respecting the semantic context of the scene. Our method involves three subnetworks: the first generate
Publikováno v:
CVPR
The task of motion transfer between a source dancer and a target person is a special case of the pose transfer problem, in which the target person changes their pose in accordance with the motions of the dancer. In this work, we propose a novel metho
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::a6386fd61c9cb277955dd19ee3d97374
Publikováno v:
ICCV
We propose a method for face de-identification that enables fully automatic video modification at high frame rates. The goal is to maximally decorrelate the identity, while having the perception (pose, illumination and expression) fixed. We achieve t