APLIKASI PENGENALAN EKSPRESI POSE JARI TANGAN MENGGUNAKAN METODE LEARNING VECTOR QUANTIZATION (LVQ) BERBASIS ANDROID
| Main Author: | KAREN DHARMAKUSUMA, NIM. 14650033 |
|---|---|
| Format: | Thesis NonPeerReviewed Book |
| Bahasa: | eng |
| Terbitan: |
, 2018
|
| Subjects: | |
| Online Access: |
http://digilib.uin-suka.ac.id/31693/2/14650033_BAB-I_IV-atau-V_DAFTAR-PUSTAKA.pdf http://digilib.uin-suka.ac.id/31693/1/14650033_BAB-II_sampai_SEBELUM-BAB-TERAKHIR.pdf http://digilib.uin-suka.ac.id/31693/ |
Daftar Isi:
- Naturally, humans have different ways of expressing expression. There are applications that offer for recognition of expressions, but most identify facial expression. In addition to the face, there are other ways of human expression, one of them through a finger pose. In the introduction of objects, used Artificial Neural Network (ANN). One method of artificial neural network is the method of Learning Vector Quantization (LVQ). Based on this, an application will be developed that can recognize the expression of the finger pose. This study aims to create an application that can recognize the expression of android-based finger poses. This ANN research uses 180 image data, in the form of 90 Training images and 90 Testing images. The training data is obtained from the edited image using adobe photoshop for background rotation and brightness, then the dimensions and sizes are resized, ie 100 pixels long and 100 pixels wide. All data will go through the process of praposesing a sobel edge detection, binaryization and feature extraction. The result of this feature extraction yields 10x10 matrix data then made into a series of one-dimensional numbers which will be used as a training process and the introduction of finger pose expression. This application has been implemented on Sony Xperia Z4 smartphone with learning rate value = 0,05 and reduce alpha = 0,1 and max epoch = 100 times and produce very good identification accuracy level, that is 81%.