Recognition of Hand Gestures Using a Smart Glove with Inductive Sensors

This study presents an improved gesture recognition system for American Sign Language (ASL) hand gestures, utilizing inductive sensors. The system integrates conductive threads to sew eight coils onto a standard glove, covering the 5 fingers, wrist, palm, and ulnar areas. Testing was conducted with 10 participants, and three machine learning algorithms (MLAs)—Random Forest (RF), Support Vector Machine (SVM), and K-Nearest Neighbors (KNN)—were evaluated for recognizing all 26 ASL letters. To increase the diversity and generalizability of the MLAs, a Generative Adversarial Network (GAN) was employed for data augmentation, generating 5050 trials per gesture. The system achieved a high accuracy of 97.46% using leave-one-subject-out cross-validation (LOSO-CV) with the RF algorithm, demonstrating effectively addresses the shortcomings of previous hand gesture recognition systems, which relied on five coils sewn onto five fingers. This proposed approach, offering a robust and practical solution for human-computer interaction.