Identifikasi Ekspresi Pada Film Animasi 3d Menggunakan Ekspresi Wajah Berbasis Neural Network dan K-Means Clustering
No Thumbnail Available
Date
2017-07-27
Authors
Wiyandari, Weni
Journal Title
Journal ISSN
Volume Title
Publisher
Politeknik Negeri Batam
Abstract
Detection of expression is something that is strategic for a company engaged
in the industrial sector in particular perfilm 3D animated movie, because it speeds
up the approval process of the animation supervisor who worked animator,
particularly in identifying expressions on character acting. To detect the expression
required an automatic system. The system requires a method of artificial
intelligence. There are two methods of artificial intelligence that is quite popular in
the identification of expression, namely Neural Networks and K-Means Clustering.
Neural Network is a supervised learning method which is well used in pattern
recognition, while the K-Means Clustering is an unsupervised learning effective
methods applied in the process of clarifying the characteristics of the object and is
not affected in the order of objects used. The experiment was conducted using both
methods to find the highest level of accuracy. In training and testing with a ratio of
80%: 20% of data using Neural Network, the epoch value 500, learning rate 0.3,
and 0.2 momentum. While at K-Means Clustering, training to use 100% of the data,
with the value itteration 500, numCluster 8 and seed 8. The test results on eight
basic expression of happy, sad, afraid, demeaning, angry, disgusted, suprised and
interestd, show the highest accuracy on Neural Network is 100%, while the
accuracy of the K-Means Clustering is 89.17%.
Description
Keywords
Teknik Informatika, Multimedia dan Jaringan, Ilmu Komputer, Ilmu Informasi, Karya Umum, Computer Sciences, Knowledge, Systems (Ilmu Komputer, Pengetahuan, Sistem), Computer Science, Computer Programming, Programs Data, Special Computer Methods (Ilmu Komputer, Pemrograman Komputer, Program Data, Metode Komputer Khusus)