ASL Recognition Based on Kinematics Derived from a Multi-Frequency RF Sensor Network
Sevgi Zübeyde Gürbüz, Ali Cafer Gürbüz, Evie Malaia, Darrin J. Griffin, Chris S. Crawford, Emre Kurtoğlu, Mohammed Mahbubur Rahman, Ridvan Aksu, Robiulhossain Mdrafi · 2020
As a means for leveraging technology in the design of Deaf spaces, this paper presents initial results on American Sign Language (ASL) recognition using RF sensing. RF sensors are non-contact, non-invasive, and protective of privacy, making them of special interest for use in personal areas. Using just the kinematic properties of signing as captured by the micro-Doppler signatures of a multi-frequency RF sensor network, this paper shows that native and imitation signing can be differentiated with %99 accuracy, while up to 20 ASL signs are recognized with an accuracy of %72 or higher.