Study of characteristics and methodology of gesture recognition system for contactless interaction with a computer using Computer Vision technologies

Abstract

The paper investigates methods of gesture recognition for contactless interaction with a computer using computer vision and machine learning technologies. Examples of application in various fields of human activity are given, where they provide greater safety, efficiency and ease of use. A methodology for recognizing hand gestures in real time is proposed. The system is based on algorithms that use image segmentation, contour detection, and hand motion classification. The main emphasis is placed on the use of OpenCV libraries for processing video streams and MediaPipe for precise tracking of finger and hand positions. The programming language used to develop the system is Python, which uses data and additional libraries to integrate various functions. A system precedent diagram was developed, the interactions of components were identified, use cases were visualized, and ways to use this diagram were proposed. The result of the study is a software product that allows you to control the mouse cursor using hand gestures. The system supports such functions as moving the cursor, left and right clicks, double left click, dragging elements, scrolling, and volume control