ROCK PAPER SCISSORS GAME WITH HAND GESTURE RECOGNITION USING MOBILENETV2 TRANSFER LEARNING

Abstract

Rock Paper Scissors (Janken/RPS) is one of the most widely played types of games across the globe. This game is usually done to draw something or just a game to determine a winner or a loser. Usually, this game is only played by two or more humans. But thanks to the times and technology, RPS can now be played between humans and computers. To realize this, a research was carried out to differentiate the shown hand object in the shape of rocks, scissors, and paper by creating a model that is capable to do so. In this case, classification performance is something that must be considered. To perform the classification, an image classifier is needed. One of the widely known classifiers is the Convolutional Neural Network (ConvNet/CNN), which is a class of neural networks typically utilized in image classification data. The human neural network is a source of inspiration in the creation of CNN. There are 3 stages that this algorithm has to use, namely the pooling layers, convolutional layers, and fully-connected layers. In this project, the final resulting model has a fairly accurate result.