Abstract: We learn to compute opticalflow by combining a classical spatial-pyramid formulation with deep learning. This estimates large motions in a coarse-to-fine approach by warping one image of a ...
Abstract: The great advancements of generative adversarial networks and face recognition models in computer vision have made it possible to swap identities on images from single sources. Although a ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results
Feedback