Hi-LASSIE: High-Fidelity Articulated Shape and Skeleton Discovery from Sparse Image Ensemble

TL;DR

Hi-LASSIE introduces a method for 3D articulated shape and skeleton reconstruction from sparse in-the-wild images without relying on large datasets or manual annotations, achieving high-fidelity results with minimal input.

Contribution

It automatically estimates class-specific skeletons and improves shape reconstructions through novel instance-specific optimization, advancing 3D reconstruction from limited images.

Findings

Achieves state-of-the-art 3D reconstructions from only 20-30 images.

Does not require user-defined shape or skeleton templates.

Outperforms previous methods in fidelity and robustness.

Abstract

Automatically estimating 3D skeleton, shape, camera viewpoints, and part articulation from sparse in-the-wild image ensembles is a severely under-constrained and challenging problem. Most prior methods rely on large-scale image datasets, dense temporal correspondence, or human annotations like camera pose, 2D keypoints, and shape templates. We propose Hi-LASSIE, which performs 3D articulated reconstruction from only 20-30 online images in the wild without any user-defined shape or skeleton templates. We follow the recent work of LASSIE that tackles a similar problem setting and make two significant advances. First, instead of relying on a manually annotated 3D skeleton, we automatically estimate a class-specific skeleton from the selected reference image. Second, we improve the shape reconstructions with novel instance-specific optimization strategies that allow reconstructions to faithful fit on each instance while preserving the class-specific priors learned across all images. Experiments on in-the-wild image ensembles show that Hi-LASSIE obtains higher fidelity state-of-the-art 3D reconstructions despite requiring minimum user input.