Best paper award and honorable mention for the Origami team at the Shape Modeling International 2023 conference

Two papers by the LIRIS Origami team received awards at the “Shape Modeling International” 2023 conference. Guillaume Gisbert received an honorable mention for his article “Inpainting holes in folded fabric meshes” with Raphaëlle Chaine and David Coeurjolly, and Mattéo Clémot received the best paper award for his article “Neural skeleton: Implicit neural representation away from the surface” with Julie Digne.

In the article “Inpainting Holes in folded meshes”, awarded an honorable mention, Guillaume Gisbert, Raphaëlle Chaine and David Coeurjolly describe an approach to filling holes in 3D scanning of the surface of a fabric when we have a mesh representation of the edge of the hole. The approach exploits the high developability of tissues to recover the intrinsic geometry of the missing patch which is then repositioned in 3D with realistic folds that perfectly match the edge of the hole.


In “Implicit neural representation away from the surface”, awarded the best paper award, Mattéo Clémot and Julie Digne propose a new regularization term for a neural implicit representation, making it possible to make the gradient field of a neural signed distance function smooth, which is particularly important far from the surface and therefore close to the median axis, which can then be extracted in a reliable and robust manner.