Year: 2023
Type: article
Abstract: A lot of attention has been paid to hand gesture detection systems in recent times due to its numerous implications and effective human-computer interaction capabilities. This paper presents an effective Hand Gesture recognition (HGR) system that is based on Deep Learning (DL). The primary goal of proposed work is to improve the classification accuracy rate while identifying different Gestures. Moreover, we have used two dataset in the proposed work, one dataset is taken from UCI machine learning repository and other was collected manually from the real world. Data Normalization technique was implemented on the original datasets in order to make the data balanced and normalized. Soon after this, a total of 20 features were extracted from the normalized data to overcome the dataset dimensionality issues. Finally, for the identification and classification of gestures, we have used an advanced variant of DL namely; Bidirectional Long Short Term Memory (Bi-LSTM) in the proposed work. The simulation of the proposed Bi-LSTM based HGR system is examined and validated by comparing it with few state of art HGR techniques using MATLAB Software. Results revealed that the proposed model achieved an accuracy of 99.876% on standard dataset and 98.366% on real time dataset.
Authors: Kumar, Davinder, Jain, Anuj, Ganesh, Aman
Language: English
Subfield: Human-Computer Interaction
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W7217404330