Paper Title: Appearance Consensus Driven Self-Supervised Human Mesh Recovery
Venue: ECCV 2020
Project Page: https://sites.google.com/view/ss-human-mesh
Paper Link: https://arxiv.org/pdf/2008.01341.pdf
Abstract: We present a self-supervised human mesh recovery framework to infer human pose and shape from monocular images in the absence of any paired supervision. Recent advances have shifted the interest towards directly regressing parameters of a parametric human model by super-vising them on large-scale datasets with 2D landmark annotations. This limits the generalizability of such approaches to operate on images from unlabeled wild environments. Acknowledging this we propose a novel appearance consensus-driven self-supervised objective. To effectively dis-entangle the foreground (FG) human we rely on image pairs depicting the same person (consistent FG) in varied pose and background (BG) which are obtained from unlabeled wild videos. The proposed FG appearance consistency objective makes use of a novel, differentiable colors-recovery module to obtain vertex colors without the need for any appearance net-work; via efficient realization of color-picking and reflectional symmetry. We achieve state-of-the-art results on the standard model-based 3D pose estimation benchmarks at comparable supervision levels. Furthermore, the resulting colored mesh prediction opens up the usage of our framework for a variety of appearance-related tasks beyond the pose and shape estimation, thus establishing our superior generalizability.