A slew of research initiatives by computer scientists in the past few decades have led to develop a wide array of computer algorithms and approaches to examine different aspects of human behavior and communication such as emotion, speech, and hand gestures. Meanwhile, most existing approaches for the recognition of hand gesture depend on the use of wearable technologies with a single sensor and can only identify a limited number of basic gestures.
Following a research initiative carried out by a research team at Comenius University a new system for hand gesture recognition driven by multiple inertial sensors, in place of the usual single sensor have been recently developed.
“The recent paper presents how multiple inertial sensors are used for hand gesture recognition,” stated one of the researchers who carried out the study. The building of custom hardware prototype and proposal of a novel transformer-based model led to demonstrate that using multiple sensors could significantly affect the accuracy of classification accuracy and allows for richer gesture vocabulary.
In fact, using WaveGlove – the newly created hand gesture recognition – the researchers were able to acquire two datasets with over 11000 hand gesture samples. Following this, the research team designed gesture vocabularies of two types, one which contains 8 whole-hand movement, and another that contains 10 more carefully designed hand gestures, with variation of finger movement.
“Meanwhile, with multiple sensors it allows richer design and classification of vocabulary gestures when compared with single handheld sensors.” The classification of gestures using multiple sensors is similar to the ones already in use in our everyday lives.” Consequently, the use of a device like WaveGlove is easier and more natural.
