NICP: Neural ICP for 3D Human Registration at Scale

arXiv:2312.14024v2 Announce Type: replace
Abstract: Aligning a template to 3D human point clouds is a long-standing problem crucial for tasks like animation, reconstruction, and enabling supervised learning pipelines. Recent data-driven methods leverage predicted surface correspondences; however, they are not robust to varied poses, identities, or noise. In contrast, industrial solutions often rely on expensive manual annotations or multi-view capturing systems. Recently, neural fields have shown promising results. Still, their purely data-driven and extrinsic nature does not incorporate any guidance toward …

abstract animation arxiv contrast cs.cv data data-driven enabling however human industrial noise pipelines registration robust scale solutions supervised learning surface tasks template type