Facial Expression Analysis using Convolutional Neural Networks
Amogh S. Gopadi1, Deepak S.2, Kiran R. B.3, Naveena A. M.4, Srividya M. S.5, Anala M. R.6
1Amogh S. Gopadi*, Dept. of Computer Science and Engineering, R.V. College of Engineering, Bengaluru, India.
2Deepak S., Dept. of Computer Science and Engineering, R.V. College of Engineering, Bengaluru, India.
3Kiran R. B., Dept. of Computer Science and Engineering, R.V. College of Engineering, Bengaluru, India.
4Naveena A. M., Dept. of Computer Science and Engineering, R.V. College of Engineering, Bengaluru, India.
5Srividya M. S., Dept. of Computer Science and Engineering, R.V. College of Engineering, Bengaluru, India.
6Anala M. R., Dept. of Computer Science and Engineering, R.V. College of Engineering, Bengaluru, India.
Manuscript received on March 30, 2020. | Revised Manuscript received on April 05, 2020. | Manuscript published on April 30, 2020. | PP: 1142-1146 | Volume-9 Issue-4, April 2020. | Retrieval Number: D8335049420/2020©BEIESP | DOI: 10.35940/ijeat.D8335.049420
Open Access | Ethics and Policies | Cite | Mendeley
© 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: Human feelings are mental conditions of sentiments that emerge immediately as opposed to cognitive exertion. Some of the basic feelings are happy, angry, neutral, sad and surprise. These internal feelings of a person are reflected on the face as Facial Expressions. This paper presents a novel methodology for Facial Expression Analysis which will aid to develop a facial expression recognition system. This system can be used in real time to classify five basic emotions. The recognition of facial expressions is important because of its applications in many domains such as artificial intelligence, security and robotics. Many different approaches can be used to overcome the problems of Facial Expression Recognition (FER) but the best suited technique for automated FER is Convolutional Neural Networks(CNN). Thus, a novel CNN architecture is proposed and a combination of multiple datasets such as FER2013, FER+, JAFFE and CK+ is used for training and testing. This helps to improve the accuracy and develop a robust real time system. The proposed methodology confers quite good results and the obtained accuracy may give encouragement and offer support to researchers to build better models for Automated Facial Expression Recognition systems.
Keywords: Convolutional Neural Network, Deep Learning, Facial Expression Recognition, OpenCV DNN.