Face Analysis Under Uncontrolled Conditions

Introduction to Part 1

Romain BELMONTE¹, Pierre TIRILLY¹, Ioan Marius BILASCO¹, Nacim IHADDADENE² and Chaabane DJERABA¹

¹University of Lille, France

²Junia ISEN, Lille, France

I1.1. Background

Over recent years, the ubiquity of sensors (i.e. smartphones, computers, in public spaces) has led to an explosion of data, especially visual data. Each day, massive amounts of visual data are produced. Every minute, 500 hours of video and 66,000 images are uploaded on YouTube and Instagram¹. Processing this amount of data is currently beyond the reach of humans and the quantity of data continues to increase. According to CISCO², video is becoming more and more important in overall traffic on the Internet³. Hence, it is critical to develop algorithms capable of understanding visual data, which is the purpose of computer vision. One of the most popular research topics within computer vision is the understanding of human behavior, since visual data represents a considerable source of information about people and their behavior. The face provides the ability to recognize people, estimate their age, gender, and their emotional state. Hence, a large amount of research is focused on the automatic analysis of facial images, which aims to extract such information. Applications cover a wide variety of domains, for example, education, transport, entertainment, security, surveillance or medicine, just to name a few. Concrete examples are engagement recognition in e-learning environments to improve ...