Bridge the gap: correlate face mask leakage and facial features with 3D morphable face models

Face masks have been proven to be effective in protecting the public against airborne transmitted diseases when fitted appropriately. However, for homemade cloth masks and surgical masks, the fit is often poor, allowing viruses to escape through the gap.

This work aims to identify the correlation between the mask leakage, mask configurations, and individual's facial features.

A novel locally morphing 3D face model, and a minimum-energy-based mask deployment model are used to systematically examine the mask fit for a large cohort of exemplars.

The results show that the mask size and tuck-in ratio, along with selective facial features, especially nose height and chin length, are key factors determining the leakage location and extent. A polynomial regression model is presented for mask fitness based on localized facial features.

This study is a complete pipeline to test various masks on a wide range of faces with controlled modification of distinct regions of the face, which is difficult to achieve with human subjects, and provide knowledge on how the masks should be designed in the future.

The face mask "fit" affects the mask's efficacy in preventing airborne transmission. To date, research on the face mask fit has been conducted mainly using experiments on limited subjects. The limited sample size in experimental studies makes it hard to reach a statistical correlation between mask fit and facial features in a population. Here, we employ a novel framework that utilizes a morphable face model and mask's deployment simulation to test mask fit for many facial characteristics and mask designs. The proposed technique is an important step toward enabling personalized mask selection with maximum efficacy for society members.