Ensemble of Multi Feature Layers in CNN for Facial Expression Recognition using Deep Learning
Chintan B. Thacker1, Ramji M. Makwana2
1Chintan B.. Thacker, Computer Engineering Department, Gujarat Technological University, Ahmedabad, India.
2Dr. Ramji M. Makwana, M.D. AIIVINE PXL Pvt. Ltd, Rajkot, Gujarat, India.
Manuscript received on November 19, 2019. | Revised Manuscript received on November 29 2019. | Manuscript published on 30 November, 2019. | PP: 9782-9787 | Volume-8 Issue-4, November 2019. | Retrieval Number: D8940118419/2019©BEIESP | DOI: 10.35940/ijrte.D8940.118419
Open Access | Ethics and 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: Facial Expression Recognition is an important undertaking for the machinery to recognize different expressive alterations in individual. Emotions have a strong relationship with our behavior. Human emotions are discrete reactions to inside or outside occasions which have some importance meaning. Involuntary sentiment detection is a process to understand the individual’s expressive state to identify his intensions from facial expression which is also a noteworthy piece of non-verbal correspondence. In this paper we propose a Framework that combines discriminative features discovered using Convolutional Neural Networks (CNN) to enhance the performance and accuracy of Facial Expression Recognition. For this we have implemented Inception V3 pre-trained architecture of CNN and then applying concatenation of intermediate layer with final layer which is further passing through fully connected layer to perform classification. We have used JAFFE (Japanese Female Facial Expression) Dataset for this purpose and Experimental results show that our proposed method shows better performance and improve the recognition accuracy.
Keywords: Convolutional Neural Networks, Facial Expression Recognition, Feature Extraction, Feature Concatenation.
Scope of the Article: Deep Learning.