Design and Development of Body Language Analysis System Based on Motion and Gesture Recognition

Abstract

This study presents the development of the Body Gesture Decoder System (BGDS), a real-time body gesture recognition system designed to improve human-computer interaction (HCI) by enabling natural, gesture-based communication. The purpose of this research is to create an accessible, lightweight system that recognizes human gestures through standard webcams, making it applicable in a variety of real-time, low-cost environments such as remote education, telemedicine, and security. The BGDS utilizes the MediaPipe Holistic framework for pose estimation, which extracts high-precision landmarks from body, face, and hand movements. A Random Forest classifier is trained on these landmarks to classify predefined gestures like- Happy, Sad, Thumbs Up, Victory. The system architecture is designed for modularity and real-time performance, achieving an accuracy of over 85% across various test conditions. It is implemented as a web-based application, making it scalable and easy to integrate with different platforms. The significance of this study lies in its demonstration of a practical, real-time solution for gesture recognition using only standard hardware, eliminating the need for expensive sensors or specialized equipment. The findings indicate that the BGDS can effectively recognize and classify a range of gestures with minimal latency, even in challenging lighting conditions. The implications of this research suggest broad potential applications in diverse sectors, including online education, telemedicine, assistive technologies, and interactive gaming. Furthermore, ethical considerations such as user privacy and data protection are central to the system’s design, ensuring that no raw video or biometric data is stored. This work paves the way for future developments in the field of body language analysis, with opportunities for expanding gesture recognition capabilities and improving system accuracy and scalability.