body2vec: 3D Point Cloud Reconstruction for Precise Anthropometry with Handheld Devices

TRUJILLO-JIMÉNEZ, MAGDA ALEXANDRA; NAVARRO, PABLO; PAZOS, BRUNO; MORALES, LEONARDO; RAMALLO, VIRGINIA; PASCHETTA, CAROLINA; DE AZEVEDO, SOLEDAD; RUDERMAN, ANAHÍ; PÉREZ, LUIS ORLANDO; DELRIEUX, CLAUDIO; GONZÁLEZ-JOSÉ, ROLANDO

Journal of Imaging

MDPI

Lugar: Basel; Año: 2020

2313-433X

Abstract: Current point cloud extraction methods based on photogrammetry generate large amountsof spurious detections that hamper useful 3D mesh reconstructions or, even worse, the possibilityof adequate measurements. Moreover, noise removal methods for point clouds are complex,slow and incapable to cope with semantic noise. In this work, we present body2vec, a model-basedbody segmentation tool that uses a specifically trained Neural Network architecture. Body2vecis capable to perform human body point cloud reconstruction from videos taken on hand-helddevices (smartphones or tablets), achieving high quality anthropometric measurements. The maincontribution of the proposed workflow is to perform a background removal step, thus avoiding thespurious points generation that is usual in photogrammetric reconstruction. A group of 60 personswere taped with a smartphone, and the corresponding point clouds were obtained automaticallywith standard photogrammetric methods. We used as a 3D silver standard the clean meshes obtainedat the same time with LiDAR sensors post-processed and noise-filtered by expert anthropologicalbiologists. Finally, we used as gold standard anthropometric measurements of the waist and hipof the same people, taken by expert anthropometrists. Applying our method to the raw videossignificantly enhanced the quality of the results of the point cloud as compared with the LiDAR-basedmesh, and of the anthropometric measurements as compared with the actual hip and waist perimetermeasured by the anthropometrists. In both contexts, the resulting quality of body2vec is equivalentto the LiDAR reconstruction.