Facial Expression for Emotion Detection using Deep Neural Networks
A. Jothimani1, Parimi Prasanth2, Shradha Anil3, J. Arun Nehru4, S. Shams Afrin Ayesha5
1A. Jothimani, Department of Computer Science and Engineering, SRM Institute of Science and Technology, Chennai (Tamil Nadu), India.
2S. Shams Afrin Ayesha, Department of Computer Science and Engineering, SRM Institute of Science and Technology, Chennai (Tamil Nadu), India.
3Parimi Prasanth, Department of Computer Science and Engineering, SRM Institute of Science and Technology, Chennai (Tamil Nadu), India.
4Shradha Anil, Department of Computer Science and Engineering, SRM Institute of Science and Technology, Chennai (Tamil Nadu), India.
5J. Arun Nehru, Department of Computer Science and Engineering, SRM Institute of Science and Technology, Chennai (Tamil Nadu), India.
Manuscript received on 20 May 2019 | Revised Manuscript received on 06 June 2019 | Manuscript Published on 15 June 2019 | PP: 242-248 | Volume-8 Issue-1S2 May 2019 | Retrieval Number: A00560581S219/2019©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: The imperative research part of the emotion recognition is the analysis of emotional state in the facial expression. The subject matter of this study is to aid the human – computer interaction more empathetic with the help of automatic emotion recognition system which will be a great step forward in the robotic field. This study proposes a novel method for the emotion detection where usage of Face detection using Haar feature-based cascade classifiers, saliency mapping and CNN architecture are implemented. The facial expression of the humans from the data set is fed into the Saliency using hyper-complex Fourier Transform (SHFT). The resulting saliency map has the extracted feature which is given as input to the CNN to perform feature modeling and output the emotional state of the human. We also exhibit that the proposed saliency model can emphasize on both minor and major salient regions more accurately than the other saliency models.
Keywords: CNN Classification, Emotion Detection, Facial Expression, Feature Extraction, Saliency Using Hyper-Complex Fourier Transform.
Scope of the Article: Deep Learning