Characterization of Facial Expression using Deep Neural Networks

Abstract

Deep learning plays a significant role in the advancement of computer vision by improving the speed and accuracy to the assigned tasks. It is opening opportunities for improvement and enhancement of processes and to initiate the human-driven tasks in an automated manner. On the basis of this growth, deep-learning algorithms are finding applications in the field CNN and RNN. The key advantage of Deep Learning algorithm is that manually extraction of features from the image is not required. The network extracts the features while training. The only input required is to provide the image to the network. The CNN’s and RNN’s have given state-of-the art results on numerous classification tasks. The Deep learning algorithm are designed for feature detection / extraction, classification and recognition of the object. The key advantage of a CNN is to remove or reduce the reliance on physics-based models, other processing methods by enabling complete learning directly from the input images of the object. The CNN and RNN together has given effective results in the area of face recognition, object recognition, scene understanding and facial expression recognition.