Deformable surface reconstruction via Riemannian metric preservation

Oriol Barbany, Adrià Colomé, Carme Torras

Abstract: Highlights•We encode a surface parametrization using a neural network for the SfT problem.•We adapt mesh dynamic constraints of a cloth simulator for shape inference.•The presented method allows for representing a continuous surface.•We analytically compute differential geometric quantities of the estimated surface.•Although based on neural networks, our method does not need offline training.