Recognition of Japanese Sign Language by Sensor-Based Data Glove Employing Machine Learning
Ji Li, Jiang Liu, Shigeru Shimamoto · 2022 IEEE 4th Global Conference on Life Sciences and Technologies (LifeTech) · 2022
This paper presents a sensor-based data acquisition glove for gesture recognition in Japanese Sign Language (JSL), which uses five flex sensors and an inertial measurement unit (IMU) to detect finger flexion and hand motion information. The detected data is sent from the Arduino Micro to a computer. We collected data from the "A" to "Ta" lines of the Japanese (kana) syllabary and using four different machine learning algorithms: Support Vector Machine (SVM), Decision Tree (DT), Random Forest (RF), and K-Nearest Neighbor (KNN) to recognize them. RF and KNN have the highest average accuracy, reaching 99.75%. Also, SVM and DT had an average accuracy of 99% and 94.25% respectively. The experimental results show that the proposed system has great potential for gesture recognition in Japanese Sign Language.