Ubiquitous Communications and Network Computing. First International Conference, UBICNET 2017, Bangalore, India, August 3-5, 2017, Proceedings

Research Article

Design of a Real-Time Human Emotion Recognition System

Download
484 downloads
  • @INPROCEEDINGS{10.1007/978-3-319-73423-1_16,
        author={D. Ashwin and Abhinav Kumar and J. Manikandan},
        title={Design of a Real-Time Human Emotion Recognition System},
        proceedings={Ubiquitous Communications and Network Computing. First International Conference, UBICNET 2017, Bangalore, India, August 3-5, 2017, Proceedings},
        proceedings_a={UBICNET},
        year={2018},
        month={1},
        keywords={},
        doi={10.1007/978-3-319-73423-1_16}
    }
    
  • D. Ashwin
    Abhinav Kumar
    J. Manikandan
    Year: 2018
    Design of a Real-Time Human Emotion Recognition System
    UBICNET
    Springer
    DOI: 10.1007/978-3-319-73423-1_16
D. Ashwin1,*, Abhinav Kumar1,*, J. Manikandan1,*
  • 1: Crucible of Research and Innovation (CORI), PES University
*Contact email: ashwindv75@gmail.com, abhinav.kumar685@gmail.com, manikandanj@pes.edu

Abstract

Emotion recognition systems are in huge demand to understand the emotion of human towards human, animals, computers, machines and systems. This has influenced such systems to be employed for applications in various domains such as website customization, study of audience reaction in theaters, gaming, software engineering, education and many more. In this paper, a real-time emotion recognition system is designed which is capable of recognizing six human emotions of any person in front of the camera, without any prior information about the person. This is achieved using a combination of both geometric and appearance based features. In order to assess and enhance the performance of the proposed design, the system is tested using standard CK+ datasets too. The system designed is evaluated using three different classifiers and their results are reported. Maximum accuracy of 98.73% is achieved by the system. The proposed system is designed using open source software and can be used for various IoT based applications too.