Detecting Hand Gestures Using Machine Learning Techniques

Abstract

The hand gesture recognition concept has recently been recognized as an essential part of the human-computer interaction (HCI) concept. Detecting and interpreting hand gestures is a very important topic. This is due to the intense desire to make communication between humans and the calculator or other device natural, away from wires, mouse, keyboards, and others. This recognition makes it possible for computers to capture and understand hand motions. Hand gestures are an important kind of nonverbal communication for a variety of reasons, including their usage in a variety of medical applications, communication between people who are hearing impaired, and robot control. Given the importance of applications for hand gesture recognition and technological progress in today's world, the purpose of the research is to shed light on the most important stage in hand gesture recognition, which is the process of detection and identifying hand gestures in the general sense; segmenting the image to obtain hand gestures before entering them into the feature extraction stages and classification. Six commonly used image segmentation methods were tested on a set of American Sign Language images in a variety of lighting conditions. When compared to the clustering and Otsu methods, the best segmenting results in terms of accuracy were obtained using the Canny and HSV color spaces.