Deep Virtual Markers for Articulated 3D Shapes

TL;DR

This paper introduces deep virtual markers, a neural network-based framework for accurately estimating dense positional information on 3D articulated models, outperforming existing methods and enabling various applications.

Contribution

It presents a novel deep virtual marker framework using sparse CNNs and soft labels for dense 3D point classification, with improved accuracy and generalizability.

Findings

Outperforms state-of-the-art on FAUST challenge

Demonstrates strong generalization to unseen data

Enables applications like non-rigid registration and texture transfer

Abstract

We propose deep virtual markers, a framework for estimating dense and accurate positional information for various types of 3D data. We design a concept and construct a framework that maps 3D points of 3D articulated models, like humans, into virtual marker labels. To realize the framework, we adopt a sparse convolutional neural network and classify 3D points of an articulated model into virtual marker labels. We propose to use soft labels for the classifier to learn rich and dense interclass relationships based on geodesic distance. To measure the localization accuracy of the virtual markers, we test FAUST challenge, and our result outperforms the state-of-the-art. We also observe outstanding performance on the generalizability test, unseen data evaluation, and different 3D data types (meshes and depth maps). We show additional applications using the estimated virtual markers, such as non-rigid registration, texture transfer, and realtime dense marker prediction from depth maps.