Pakistani Phrasal Sign Language Classification using Surface Electromyography
Muhammad Umar Khan, Sumair Aziz, Syed Zohaib Hassan Naqvi, Fatima Amjad, Maheen Shakeel · 2020 International Conference on Computing and Information Technology (ICCIT-1441) · 2020
Electromyogram (EMG) is the electrical measure of the muscular activity which can vary depending upon motion is being performed. This varied muscular activity is picked by the EMG which can be used to distinguish different muscular activity. Electromyogram (EMG) is the electrical measure of the muscular activity which can vary depending upon what motion is being performed. This phenomenon forms the basis of this research which is the detection of various hand gestures that are performed during sign language communication. Pakistani sign language (PSL) data were acquired using the BIOPAC system. A total of 550 signals were collected. The signals were preprocessed using Empirical mode decomposition, which is succeeded by feature extraction giving feature vector consisting of statistical, spectral, time domain, and Local Ternary Pattern (LTP). These features were fed to support vector machines, which gives optimized results of 85.4% of accuracy, 85.36% of sensitivity, and 85.81 of specificity.