AI Implementations of Hand Gestures Detection in The “Pong Ball” Game Using The Convolutional Neural Network Algorithm

Abstract

Keyboard, mouse, and joystick are hardware devices that are often used for interaction between humans and computers which are mechanical. Although accurate, the model of the interaction method is not natural as humans interact with each other. To be able to operate, it is necessary to have direct contact between the user and the computer. The thing studied here is the pattern of hand movements. This was investigated by using a program with the help of a webcam camera to perform hand object recognition. hand motion pattern detection is implemented using the cvzone library and the Convolutional Neural Network algorithm. The application of the accuracy of the program data made is by playing the game "Pong Ball". Trial using a simple “Pong Ball” game where the movement is up and down. When the hand is moved up, the bat pong ball will go up according to the pattern of motion performed, and vice versa. From research trials with bright lighting, the game "Pong Ball" runs smoothly, while when the lighting received is less or dim, the motion sensor will automatically slow down. Not only that, the program is also able to detect 2 to 4 hand objects that allow the game between teams to occur.