Facial Emotion Recognition Using Deep Cnn Based Features
Jyostna Devi Bodapati1, N. Veeranjaneyulu2
1Jyostna Devi Bodapati, Department of CSE, Vignan’s Foundation for Science, Technology and Research, Vadlamudi, Guntur (A.P), India.
2N. Veeranjaneyulu, Department of IT, Vignan’s Foundation for Science, Technology and Research, Vadlamudi, Guntur (A.P), India.
Manuscript received on 01 May 2019 | Revised Manuscript received on 15 May 2019 | Manuscript published on 30 May 2019 | PP: 1928-1931 | Volume-8 Issue-7, May 2019 | Retrieval Number: G6256058719/19©BEIESP
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: The objective of emotion recognition is identifying emotions of a human. The emotion can be captured either from face or from verbal communication. In this work we focus on identifying human emotion from facial expressions. Facial emotion recognition is one of the useful task and can be used as a base for many real-time applications. It can be used as a part of many interesting and useful applications like Monitoring security, treating patients in medical field, marketing research, E-learning etc;. We humans can easily identify the emotion of other humans without any effort. Automatic detection of emotion of a human face is important due to its use in real-time applications. The recent advance in GPU has taken many applications like face recognition, hand written digit recognition and object recognition to the next level. Especially the pretrained CNN based features better represent the images. Pretrained CNN features represent the most discriminative features and hence allows for better performance. Feature representation plays a major role on the performance of any machine learning algorithm. After observing unbelievable performance with deep learning models, we propose to use the deep convolutional features to better represent the given image instead of using the traditional handcrafted features. The downside of the deep learning models is that they require large datasets to obtain better performance. To leverage the use of deep learning models without the requirement of large datasets is to use pre-trained models. For feature extraction pre-trained Convolutional Neural Networks model (VGG16) is used and the concept of Deep Neural Networks model is used for classification. To show the performance of the proposed model, Extended Cohn-Kanade (CK+) benchmark dataset is used for the experimental studies. Based on the experimental results we claim that these unsupervised features better represent the images compared the handcrafted features.
Keyword: Facial Emotion Recognition, CNN, RBF Kernels, Extended Cohn-Kanade, Multi-class SVM.
Scope of the Article: Deep Learning.