Accuracy Improvement of Object Selection in Gaze Gesture Application using Deep Learning
Penulis/Author
M. ALFAROBY (1); Dr.Eng. Ir. Sunu Wibirama, S.T., M.Eng., IPM. (2); Dr.Eng. Ir. Igi Ardiyanto, S.T., M.Eng. (3)
Tanggal/Date
1 2020
Kata Kunci/Keyword
Abstrak/Abstract
Gaze-based interaction is a crucial researcharea. Gaze gesture provides faster interaction between a userand a computer application because people naturally look atthe object of interest before taking any other actions. Spon-taneous gaze-gesture-based application uses gaze-gesture asan input modality without performing any calibration. Theconventional eye tracking systems have a problem with lowaccuracy. In general, data captured by eye tracker containserrors and noise within gaze position signal. The errorsand noise affect the performance of object selection in gazegesture based application that controls digital contents onthe display using smooth-pursuit eye movement. The con-ventional object selection method suffers from low accuracy(<80%). In this paper, we addressed this accuracy problemwith a novel approach using deep learning. We exploiteddeep learning power to recognize the pattern of eye-gazedata. Long Short Term Memory (LSTM) is a deep learningarchitecture based on recurrent neural network (RNN). Weused LSTM to perform object selection task. The datasetconsisted of 34 participants taken from previous study ofobject selection technique of gaze gesture-based application.Our experimental results show that the proposed methodachieved 96.17% of accuracy. In future, our result may beused as a guidance for developing gaze gesture application
Level
Internasional
Status
Dokumen Karya
No
Judul
Tipe Dokumen
Aksi
1
Accuracy Improvement of Object Selection inGaze Gesture Application using Deep Learning.pdf
[PAK] Full Dokumen
2
2020_ICITEE_Sunu_Artikel_dan_Sertifikat.pdf
Artikel dan Sertifikat/Bukti Kehadiran/Pasport (jika tidak ada sertifikat)
