CD Skripsi
Identifikasi Citra Motif Permainan Tadisional Kartu Koa Menggunakan Convolutional Neural Network (Cnn)
ABSTRACT
Koa card game is a traditional card game that is popular in Minangkabau. This koa card game is played by 4 people, each player will get 11 cards each, to get the winner of this game when one of the players manages to collect the same koa card motif from the cards that are dealt. This game is usually played as entertainment for the community which is usually played in coffee shops or played when there are traditional events, or when one of the residents is holding a wedding reception. However, as time goes by and technology has resulted in games being carried out virtually. To be able to play this game virtually, this study aims to build an image identification model to distinguish koa card motifs by using one of the Deep Learning algorithms to identify an image, namely the Convolutional Neural Network. The results of this study use a dataset of 5100 images requiring 4convolutional layer and maxpooling, and 2 fully connected layer, as well as used batch size 16 and epoch 50 which produces the highest accuracy in the 43rd epoch of 96.88% in the training process, and on testing preoces uses 90 new data which results in 98% accuracy.
Keywords: Convolutional Neural Network, Image Identification, Koa
Tidak tersedia versi lain