FAUST: Dataset and Evaluation for 3D Mesh Registration

TLDR

New scanning technologies increase the importance of 3D mesh data, demanding reliable alignment for building full models, statistical shape models, retrieval, and tracking, especially for non‑rigid articulated objects like human bodies, yet real‑world data lack ground‑truth correspondences present in synthetic datasets. The authors introduce FAUST, a dataset of 300 scans of 10 people in varied poses, and a novel mesh registration technique that fuses shape and appearance information to produce high‑quality alignments. The method employs high‑frequency textured scans and an extensive validation pipeline to establish accurate ground truth, combining shape and appearance cues for registration. Current shape registration methods struggle on the real‑world FAUST data, and the dataset with an evaluation website is publicly available.

Abstract

New scanning technologies are increasing the importance of 3D mesh data and the need for algorithms that can reliably align it. Surface registration is important for building full 3D models from partial scans, creating statistical shape models, shape retrieval, and tracking. The problem is particularly challenging for non-rigid and articulated objects like human bodies. While the challenges of real-world data registration are not present in existing synthetic datasets, establishing ground-truth correspondences for real 3D scans is difficult. We address this with a novel mesh registration technique that combines 3D shape and appearance information to produce high-quality alignments. We define a new dataset called FAUST that contains 300 scans of 10 people in a wide range of poses together with an evaluation methodology. To achieve accurate registration, we paint the subjects with high-frequency textures and use an extensive validation process to ensure accurate ground truth. We find that current shape registration methods have trouble with this real-world data. The dataset and evaluation website are available for research purposes at http://faust.is.tue.mpg.de.