Reconstruction of People in Loose Clothing

TL;DR

This paper introduces SHARP, an end-to-end neural network that reconstructs detailed 3D human models in loose clothing from a single image by combining parametric body priors with peeled depth maps.

Contribution

The paper presents a novel fusion method of parametric and non-parametric representations for accurate 3D human reconstruction in loose clothing from monocular images.

Findings

SHARP outperforms existing methods on Cloth3D and THuman datasets.

It accurately recovers detailed geometry and appearance of clothed humans.

The approach is efficient and retains geometrically consistent body parts.

Abstract

3D human body reconstruction from monocular images is an interesting and ill-posed problem in computer vision with wider applications in multiple domains. In this paper, we propose SHARP, a novel end-to-end trainable network that accurately recovers the detailed geometry and appearance of 3D people in loose clothing from a monocular image. We propose a sparse and efficient fusion of a parametric body prior with a non-parametric peeled depth map representation of clothed models. The parametric body prior constraints our model in two ways: first, the network retains geometrically consistent body parts that are not occluded by clothing, and second, it provides a body shape context that improves prediction of the peeled depth maps. This enables SHARP to recover fine-grained 3D geometrical details with just L1 losses on the 2D maps, given an input image. We evaluate SHARP on publicly available Cloth3D and THuman datasets and report superior performance to state-of-the-art approaches.