Gesture Recognition Using MediaPipe for Online Realtime Gameplay

Abstract

Hand gesture recognition has advanced greatly in the recent years due to its effectiveness in interacting with computers, machines, and other equipment and devices. It has been used in various fields such as hospitals, sign language recognition. While applications of the techniques for traditional video games at this current stage are still quite limited, we see the potentials using gesture-based controls as training and rehabilitation tools. This paper explores in particular how gestures can be used to play video games. It uses the gestures captured via the user’s video camera and performs various actions in the game. Two games, Hill Climb Racing and Subway Surfers, have been investigated in this paper. For Hill Climb Racing, only two hand gestures are needed since there are only two actions in the game. Body gestures and movements are used to play the game subway surfers. Promising results (real time with webcam only) indicating practicability of the approach are obtained.