Gesture Control Mouse

EasyChair Preprint 12554

Abstract

The evolution of computer commerce towards greater intuitiveness has a rich history, with gesture-based communication being one of the most natural modes for mortal beings. However, vision-based hand gesture recognition presents a formidable challenge due to the intricate computations involved, stemming from the complex degrees of freedom inherent in mortal hands. In this paper, we propose leveraging hand gestures captured via webcam instead of traditional mice input to facilitate natural and intuitive mortal-computer interaction. A skin-finding system is employed to generate segmented hand images, effectively isolating them from the background. Furthermore, we utilize outline and convex hull algorithms to delineate the hand area and ascertain the number of fingertip positions in the gesture image, crucial for interaction with onscreen buttons. Additionally, a method for detecting hand gesture dynamics is proposed. The results demonstrate the efficacy of this system, affirming its capability to enable seamless computer interaction via hand gestures, thus obviating the need for conventional mouse input.

Keyphrases: MediaPipe, OpenCV, Python, Virtual Mouse, machine learning

@booklet{EasyChair:12554,
  author    = {Madhukar Reddy Visavarapu and Suraj Reddy Vangala and Sai Lohith Vadada and Gowrav Vallabhaneni and Ankitha Gandhi},
  title     = {Gesture Control Mouse},
  howpublished = {EasyChair Preprint 12554},
  year      = {EasyChair, 2024}}