Skip to content
Jorge MF edited this page Sep 10, 2019 · 2 revisions

FSGAN: Subject Agnostic Face Swapping and Reenactment (Agu 2019)
Set of techniques and models to face swapping in videos using GANs. One model to detect face landmarks, another to interpolate the best image for the current frame, one for detecting face and hair and another one to blend the face into the video.

4D Spatio-Temporal ConvNets: Minkowski Convolutional Neural Networks [code] (Apr 2019)
Sparse 4D convolution to process temporal 3D input. Used for segmentation in a 3D dataset of rooms.

Clone this wiki locally