Human Face Recognition

Face recognition is an extremely challenging task, and it is one that humans are remarkably adept at. It takes us mere milliseconds to recognize a face of someone we know. It’s so natural and evolutionarily beneficial that we rarely stop to think just how difficult of a task it actually is.

mechanisms of face recognition

Face recognition shares mechanisms with other types of visual recognition such as object and scene recognition, while others are specific to recognizing faces.

what is easy? what is hard?

A wide variety of studies explored the limits and challenges of human face recognition. It is easy for us to recognize an upright face but a lot harder to recognize a face that is upside down. Additionally, we perform poorly when asked to recognize a person based on some of their features (e.g., top half of their face.)

stages of human face recognition

According to Tsao&Livingstone, faces recognition face be divided into three stages: detection, measurement/categorization, and identification

individual differences

People differ vastly in their ability to recognize faces, from prosopagnosiacs to superrecognizers. Similarly, some faces are a lot easier to recognize than others

Learn More!

Computer Face Recognition

How can computer recognize faces automatically?

What is computer's "bottom-up stimulus"?

Pixels! And they are just numbers!

Face Detection

Computers have to figure which pixels are faces and which are noise.

Feature Extraction

This is the procedure of extracting relevant information from a face image to retain the most valuable knowledge about the face while keeping size of information manageable to process by the computer.

Face Classification/Recognition

With the labels of a set known faces, computers use a variety of approaches to determine which person a new face belongs to.

Learn More!

Challenges

There are a lot of similarities between the processes of human and computer face recognition which makes a lot of sense given that computer face recognition is informed by the way humans do it. So a lot of challenges faced by humans will also apply to computers. But there are also several noteworthy differences.

Human

Orientation

We are used to faces being presented to us in a very specific way - that is, upright. It has been repeatedly shown that people's face recognition ability suffers greatly when the face is presented upside down. For a computer, however, that would make no difference since the relationship between pixels in the image will remain the same.

Computer

Optical technology

A face recognition system should be aware of the format in which the input images are provided. There are different cameras, with different features, different weaknesses and problems. Usually, most of the recognition processes involve a preprocessing step that deals with this problem.

Expression

Facial expression is another variability provider. Several algorithms don’t deal with this problem in a explicit way, but they show a good performance when different facial expressions are present. On the other hand, the addition of expression variability to pose and illumination problems can become a real impediment for accurate face recognition.

Algorithm concern

Speed, Memory Usage, Scalability, Automatism, Portability.

Both

Visual similarity

You may argue that people you know well look nothing alike but in reality we all have roughly the same facial features that are organized in the same way.

Lighting conditions and angles

We encounter faces in various lighting conditions and from different angles which means that different information about a face is available to us every time.

Occlusion & Constant Change

People’s faces constantly change: people age, apply makeup, wear hats and beards, and yet we usually have no trouble recognizing a person we have not seen in years.

References

  • Brewer, K. (2005). The psychology of face recognition: A brief introduction. Grays: Orsett Psychological Services.
  • De Carrera, P. F., & Marques, I. (2010). Face recognition algorithms. Master's thesis in Computer Science, Universidad Euskal Herriko.
  • Ellis, H. D., Shepherd, J. W., & Davies, G. M. (1979). Identification of familiar and unfamiliar faces from internal and external features: Some implications for theories of face recognition. Perception, 8(4), 431-439.
  • Geitgey, A. (2016, July 24). Machine Learning is Fun! Retrieved from https://medium.com/@ageitgey/machine-learning-is-fun-part-4-modern-face-recognition-with-deep-learning-c3cffc121d78
  • Farah, M. J., Wilson, K. D., Drain, M., & Tanaka, J. N. (1998). What is" special" about face perception?. Psychological review, 105(3), 482.
  • Moscovitch, M., Winocur, G., & Behrmann, M. (1997). What is special about face recognition? Nineteen experiments on a person with visual object agnosia and dyslexia but normal face recognition. Journal of cognitive neuroscience, 9(5), 555-604.
  • Rezlescu, C., Pitcher, D., & Duchaine, B. (2012). Acquired prosopagnosia with spared within-class object recognition but impaired recognition of degraded basic-level objects. Cognitive Neuropsychology, 29(4), 325-347.
  • Tsao, D. Y., & Livingstone, M. S. (2008). Mechanisms of face perception. Annu. Rev. Neurosci., 31, 411-437.
  • Valentine, T. (1991). A unified account of the effects of distinctiveness, inversion, and race in face recognition. The Quarterly Journal of Experimental Psychology Section A, 43(2), 161-204.
  • Wilmer, J. B., Germine, L., Chabris, C. F., Chatterjee, G., Williams, M., Loken, E., ... & Duchaine, B. (2010). Human face recognition ability is specific and highly heritable. Proceedings of the National Academy of sciences, 107(11), 5238-5241.
  • Young, A. W., Hellawell, D., & Hay, D. C. (2013). Configurational information in face perception. Perception, 42(11), 1166-1178.