Conference paper

Face Emotional Detection using Computational Intelligence based Ubiquitous Computing

Irawan Dwi Wahyono, Djoko Saryono, Muhammad Ashar, Khoirudin Asfani, Sunarti

Proceedings - 2019 International Seminar on Application for Technology of Information and Communication: Industry 4.0: Retrospect, Prospect, and Challenges, iSemantic 2019 · 2019

16Citations
5Authors

Summary

Summary Face Emotional Detection using Computational Intelligence based Ubiquitous Computing
Key quote

The system achieved an accuracy of 77.5% in detecting face emotions, with an error rate of 22.5%.

Wahyono et al., 2019

Details

Abstract

The ubiquitous use of industrial technology 4.0 is increasing mainly to facilitate human work. This is because ubiquitous is a small and inexpensive device but has limitations in resources. Research on ubiquitous continues to be developed especially in the use of sensors in ubiquitous. Optimization of face sensors for face detection has also been done. High computing for face detection optimization causes overload on ubiquitous and low accuracy of face detection. This is because resources are limited to ubiquitous. This study developed the human face emotional detection using Computational Intelligence in ubiquitous devices. The computations carried out on ubiquitous devices to detect facial emotions using the K-Means algorithm to segment colors in the results of face images. Photos of segmentation were carried out by histogram to get the gray value. Then, the value is matched with the dataset in the database. The dataset has been classified based on 3 face emotional conditions which are happy, normal, and sad. The dataset classification uses the Nearest Neighbors algorithm. Results of the system testing are done by dataset testing and direct testing on the application. The test results were lined up with an average accuracy of 77.5% with an error of 22.5%. © 2019 IEEE.

Affiliations

Universitas Negeri Malang, Malang, Jawa Timur, Indonesia

Institute of Electrical and Electronics Engineers Inc.