Face Emotion Recognition From Static Image Based on Convolution Neural Networks

Human-Machine Interaction systems have not yet reached all the emotional and social capacities. In this paper, we propose a face emotion recognition system from static image based on the Xception convolution neural network architecture and the K-fold-cross-validation strategy. The proposed system was improved using the fine-tuning method. The Xception model pre-trained on ImageNet database for objects recognition was fine-tuned to recognize seven emotional states. The proposed system is evaluated on the database recorded during the Empathic project and the AffectNet database. Our experimental results achieve an accuracy of 62%, 69% on Empathic and AffectNet databases respectively using the fine-tuning strategy. Combined the AffectNet and Empathic databases to train our proposed model, show significant improvement in the emotion recognition that achieves an accuracy of 91.2% on Empathic database.

I. Introduction

Emotions are essential in our life, it plays a main role in all our social interactions and they make it possible to improve communication between persons, and allows ensure a better understanding of the message conveyed. These different roles interest many applications that try to automate the recognition of emotions. These applications are related to many areas, such as marketing applications for example, measure customer satisfaction and predict the products that interest them, in the medical field for example, detection of certain psychological illnesses, help in learning emotions for autistic children, and in the security field for example, stress detection. Automatic emotion detection system are also used in the Human-Machine interaction application. Recently the Human-Machine Interaction systems are more used in different applications such as in smart phone, self drive car and virtual coach. Most of these systems have not yet reached all the emotional and social capacities necessary for more real and natural interaction. In fact, the recognition of the emotional state of individuals makes it possible to better manage the interactivity between the system and the different users by modifying the behavior according to the state of the users. Emotion detection is the process of identifying the emotions, usually from facial and verbal expressions. In this work we focus on the face emotion detection task.