Vector Machine Learning Models for Human Gesture Recognition

Authors- Pham Quoc Thang, Hoang Thi Lam

Abstract-Human gesture recognition is a complex yet critical task in the field of computer vision, driven by advancements in motion-capture technology and the availability of devices like Microsoft’s Kinect sensor. This paper explores the application of vector machine learning models, specifically Support Vector Machines (SVM), Simplified Support Vector Machines (SimpSVM), and Relevance Vector Machines (RVM), to the problem of human gesture recognition. Our experiments on the Microsoft Research Cambridge-12 Kinect dataset demonstrate that these vector machine learning models achieve high accuracy and competitive performance in gesture classification. SVM and SimpSVM, in particular, exhibit superior accuracy compared to RVM, though RVM shows advantages in classification speed due to fewer support vectors. This study confirms that vector machine learning models are effective for human gesture recognition, providing a promising direction for future research and application in interactive systems.

DOI: /10.61463/ijset.vol.12.issue3.187