公开(公告)号：US11783627B2

公开(公告)日：2023-10-10

申请号：US17163410

申请日：2021-01-30

Applicant: Massachusetts Institute of Technology

Inventor： Canan Dagdeviren ,  Farita Tasnim

IPC: G06V40/16 ,  G06F3/01 ,  G06N20/00

CPC classification number: G06V40/166 ,  G06F3/012 ,  G06F3/017 ,  G06N20/00 ,  G06V40/172 ,  G06V40/174

Abstract: A conformable sensor module may conform to skin of a user's face. The sensor module may include multiple piezoelectric strain sensors. The sensor module may measure mechanical strain of facial skin that occurs while the user makes facial gestures. To do so, the sensor module may take a time series of multiple measurements of strain of the user's facial skin at each of multiple locations on the user's face, while the user makes a facial gesture. The resulting spatiotemporal data regarding facial strain may be fed as an input into a trained machine learning algorithm. The trained machine learning algorithm may, based on this input, classify a facial gesture. A computer may determine content associated with the classification. The content may be outputted in audible or visual format. This may facilitate communication by patients with neuromuscular disorders who are unable to vocalize intelligible speech.

公开(公告)号：US20210256246A1

公开(公告)日：2021-08-19

申请号：US17163410

申请日：2021-01-30

Applicant: Massachusetts Institute of Technology

Inventor： Canan Dagdeviren ,  Farita Tasnim

IPC: G06K9/00 ,  G06F3/01 ,  G06N20/00

Abstract: A conformable sensor module may conform to skin of a user's face. The sensor module may include multiple piezoelectric strain sensors. The sensor module may measure mechanical strain of facial skin that occurs while the user makes facial gestures. To do so, the sensor module may take a time series of multiple measurements of strain of the user's facial skin at each of multiple locations on the user's face, while the user makes a facial gesture. The resulting spatiotemporal data regarding facial strain may be fed as an input into a trained machine learning algorithm. The trained machine learning algorithm may, based on this input, classify a facial gesture. A computer may determine content associated with the classification. The content may be outputted in audible or visual format. This may facilitate communication by patients with neuromuscular disorders who are unable to vocalize intelligible speech.