Facial and expression recognition for the blind using computer vision

View/ Open
Date
2012Author
Astler, Douglas
Chau, Harrison
Hsu, Kailin
Hua, Alvin
Kannan, Andrew
Lei, Lydia
Nathanson, Melissa
Paryavi, Esmaeel
Rosen, Michelle
Unno, Hayato
Wang, Carol
Zaidi, Khadija
Zhang, Xuemin
Advisor
Chellappa, Rama
Tang, Cha-Min
Metadata
Show full item recordAbstract
The majority of communication between humans is comprised of nonverbal cues. Visual cues, such as expressions and nodding, are not readily accessible to the blind. Our team has developed an assistive device based on computer vision, which relays facial recognition and expression information to a blind user and is designed to store images of people the user frequently interacts with and analyze faces for expressions. The control a user has over his/her surroundings
while receiving real-time feedback contributes to a unique device. In order to design a device suited best to the blind user's needs, we engaged sighted and blind
participants in surveys and interviews to understand their views and preferences
regarding the availability of a computer vision system that, in real-time, can pro-
vide information about identity and expressions of humans. This thesis discusses
the development methodology, the selection of algorithms for recognizing faces
and expressions and physical designs of the device, and the results on subject
tests with blind participants to gauge the e ectiveness of our design.
Notes
Gemstone Team FACE