Persian sign language recognition using IMU and surface EMG sensors

A sign language recognition (SLR) system has been broadly used by deaf individuals as a communicative tool. Progress of SLR systems paves the way for the development of Human-computer interaction (HCI) since sign language is the most structured form, and every movement has a specific meaning. In this paper, a wearable low-cost device based on surface electromyography (sEMG) and Inertial Measurement Unit (IMU) sensors was designed and fabricated. The fusion of these two sensors will improve the system's accuracy to capture signs. In this work, sEMG and IMU recordings were collected from ten volunteers while performing 20 commonly used Persian Sign Language (PSL) signs ten times in defined time gaps. To make the proposed algorithm computationally efficient, the 25 highest-ranked features of two modalities (sEMG, IMU) were extracted and classified by the KNN classifier that achieved 96.13% average accuracy. These results demonstrate the feasibility of our purposed method for PSL recognition.

Automated summary

A wearable device using sEMG and IMU sensors was designed to improve accuracy in sign language recognition. Through experiments on PSL signs and efficient classification methods, the proposed system demonstrated a high average accuracy of 96.13%.