DPHMs: Diffusion Parametric Head Models for Depth-based Tracking

arXiv:2312.01068v2 Announce Type: replace
Abstract: We introduce Diffusion Parametric Head Models (DPHMs), a generative model that enables robust volumetric head reconstruction and tracking from monocular depth sequences. While recent volumetric head models, such as NPHMs, can now excel in representing high-fidelity head geometries, tracking and reconstructing heads from real-world single-view depth sequences remains very challenging, as the fitting to partial and noisy observations is underconstrained. To tackle these challenges, we propose a latent diffusion-based prior to regularize volumetric head reconstruction …

abstract arxiv cs.cv diffusion excel fidelity generative head parametric robust tracking type view world