The Application of Deep Learning in Micro-Expression Recognition

Abstract

With the advent of the era of big data, the application of artificial intelligence and big data technology has led to widespread interest in facial expression recognition. In addition to the common macro-expressions in daily life, facial expressions also have an imperceptible subtle expression called micro-expressions. Micro-expression is a very fast expression, lasting only 1/25 seconds to 1/5 seconds, expressing the real emotions that people try to suppress and hide. Micro-expressions have important applications in public safety, judicial criminal investigation, clinical medicine, etc. Therefore, the research on micro-expression recognition has been increasing in recent years. In the year of 2006, Hinton's proposal of deep learning in an article in «Science» made it appear in front of the world as a new field of machine learning, which set off a wave of deep learning. Deep learning is currently one of the hottest research directions in artificial intelligence and machine learning, and it is widely used in speech recognition, natural language processing, image recognition and other fields. Because of the characteristics of micro-expressions such as short time and subtle changes, traditional machine learning algorithms have poor robustness. Therefore, this paper summarizes the research on micro-expression recognition based on deep learning methods, mainly including some mainstream algorithms such as DBN, CNN, and discusses the development problems and trends of deep learning on micro-expression recognition, hoping to provide reference for the subsequent research in this field.