Automatic recognition of the American sign language ﬁngerspelling alphabet to assist people living with speech or hearing impairments
Quesada Quirós, Luis
López Herrera, Gustavo
Guerrero Blanco, Luis Alberto
MetadataShow full item record
Sign languages are natural languages used mostly by deaf and hard of hearing people. Different development opportunities for people with these disabilities are limited because of communication problems. The advances in technology to recognize signs and gestures will make computer supported interpretation of sign languages possible. There are more than 137 different sign languages around the world; therefore, a system that interprets them could be beneficial to all, especially to the Deaf Community. This paper presents a system based on hand tracking devices (Leap Motion and Intel RealSense), used for signs recognition. The system uses a Support Vector Machine for sign classification. Different evaluations of the system were performed with over 50 individuals; and remarkable recognition accuracy was achieved with selected signs (100% accuracy was achieved recognizing some signs). Furthermore, an exploration on the Leap Motion and the Intel RealSense potential as a hand tracking devices for sign language recognition using the American Sign Language fingerspelling alphabet was performed.