3DBODY.TECH 2021 - Paper 21.40

S. L. Sokolowski et al., "Exemplar 3D Faces and N95 Pleated Mask Measurement by Sex and Race", Proc. of 3DBODY.TECH 2021 - 12th Int. Conf. and Exh. on 3D Body Scanning and Processing Technologies, Lugano, Switzerland, 19-20 Oct. 2021, #40, https://doi.org/10.15221/21.40.

Title:

Exemplar 3D Faces and N95 Pleated Mask Measurement by Sex and Race

Authors:

Susan L. SOKOLOWSKI 1, Jacob A. SEARCY 2, Daniel CALABRESE 1, Yu ZOU 1

1 Sports Product Design - University of Oregon, Portland (OR), USA;
2 Data Science - University of Oregon, Eugene (OR), USA

Abstract:

The National Institute for Occupational Safety and Health (NIOSH) requires N95 masks to provide adequate fit to at least 95% of the US population through a fit test panel, with defined face dimensions. However, researchers for many years have reported discrepancies with N95 mask fit for women and minorities. More recently with the COVID-19 pandemic, these issues were critically raised again locally when a hospital mask fitter explained to the PI issues of mask fit with women and minorities, and the challenges of finding available masks on the market that protect appropriately. A pilot study was conducted with 3D face scans from the Civilian American and European Surface Anthropometry Resource (CAESAR) dataset to uncover how exemplar faces identified through an unsupervised machine learning algorithm measured and compared to an existing N95 pleated mask (Crosstex Isolator Plus). The algorithm was based on a Variational Autoencoder (VAE) with a Point-Net inspired encoder and decoder architecture trained on Human point-cloud data obtained from the CEASAR dataset. The pilot demonstrated that the algorithm worked well to identify exemplars for sizing buckets based on sex and race and proved that ML could help replace tedious anthropometric measuring practices to develop sizing systems. From the anthropometric measures collected from the exemplars, 37.5% fitted lengthwise and widthwise into the mask, where 16.7% of the exemplars fitted in the length only, and 20.8% in the width only. Twenty-five percent did not fit at all into the mask. The results of this work highlight how critical it is for N95 mask manufacturers to look at the sizing and fit of masks differently. Multiple sizes are needed within a mask style and sex/race must be considered through relevant 3D anthropometric face/head data that represents users appropriately when developing equitable sizing systems for Personal Protective Equipment (PPE).

Keywords:

3D face scans, Machine learning, N95 Masks, Sizing

Details:

Full paper: 2140sokolowski.pdf
Proceedings: 3DBODY.TECH 2021, 19-20 Oct. 2021, Lugano, Switzerland
Paper id#: 40
DOI: 10.15221/21.40
Presentation video: 3DBodyTech2021_40_Sokolowski.mp4

Copyright notice

© Hometrica Consulting - Dr. Nicola D'Apuzzo, Switzerland, hometrica.ch.
Reproduction of the proceedings or any parts thereof (excluding short quotations for the use in the preparation of reviews and technical and scientific papers) may be made only after obtaining the specific approval of the publisher. The papers appearing in the proceedings reflect the author's opinions. Their inclusion in these publications does not necessary constitute endorsement by the editor or by the publisher. Authors retain all rights to individual papers.


Note: click the + on the top left of the page to open/close the menu.