Abstract
We compared face identification by humans and machines using images taken under a variety of uncontrolled illumination conditions in both indoor and outdoor settings. Natural variations in a person's day-to-day appearance (e.g., hair style, facial expression, hats, glasses, etc.) contributed to the difficulty of the task. Both humans and machines matched the identity of people (same or different) in pairs of frontal view face images. The degree of difficulty introduced by photometric and appearance-based variability was estimated using a face recognition algorithm created by fusing three top-performing algorithms from a recent international competition. The algorithm computed similarity scores for a constant set of same-identity and different-identity pairings from multiple images. Image pairs were assigned to good, moderate, and poor accuracy groups by ranking the similarity scores for each identity pairing, and dividing these rankings into three strata. This procedure isolated the role of photometric variables from the effects of the distinctiveness of particular identities. Algorithm performance for these constant identity pairings varied dramatically across the groups. In a series of experiments, humans matched image pairs from the good, moderate, and poor conditions, rating the likelihood that the images were of the same person (1: sure same - 5: sure different). Algorithms were more accurate than humans in the good and moderate conditions, but were comparable to humans in the poor accuracy condition. To date, these are the most variable illumination- and appearance-based recognition conditions on which humans and machines have been compared. The finding that machines were never less accurate than humans on these challenging frontal images suggests that face recognition systems may be ready for applications with comparable difficulty. We speculate that the superiority of algorithms over humans in the less challenging conditions may be due to the algorithms' use of detailed, view-specific identity information. Humans may consider this information less important due to its limited potential for robust generalization in suboptimal viewing conditions.
- Adini, Y., Moses, Y., and Ullman, S. 1997. Face recognition: The problem of compensating for changes in illumination direction. IEEE Trans. Pattern Anal. Mach. Intell. 19, 721--732. Google ScholarDigital Library
- Braje, W. 2003. Illumination encoding in face recognition: Effect of position shift. J. Vision 3, 161--170.Google ScholarCross Ref
- Braje, W., Kersten, D., Tarr, M. J., and Troje, N. 1999. Illumination effects in face recognition. Psychobiology 26, 371--380.Google Scholar
- Burton, A. M., Jenkins, R., and Schweinberger, S. R. 2011. Mental representations of familiar faces. British J. Psychol.102, 943--958.Google ScholarCross Ref
- Gobbini, M. I. and Haxby, J. V. 2011. Neural systems for recognition of familiar faces. Neuropsychologica 45, 32--41.Google ScholarCross Ref
- Gross, R., Baker, S., Matthews, I., and Kanade, T. 2005. Face recognition across pose and illumination. In Handbook of Face Recognition, S. Z. Li and A. K. Jain, Eds., Springer, Berlin, 193--216.Google Scholar
- Hancock, P. J. B., Bruce, V., and Burton, A. M. 2000. Recognition of unfamiliar faces. Trends Cognitive Sci. 4, 330--337.Google ScholarCross Ref
- Haxby, J., Hoffman, E., and Gobbini, M. 2000. The distributed human neural system for face perception. Trends Cognitive Sci. 20, 6, 223--233.Google ScholarCross Ref
- Hill, H. and Bruce, V. 1996. Effects of lighting on the perception of facial surface. J. Experiment. Psychol. 22, 986--1004.Google Scholar
- Jenkins, R., White, D., Monfort, X. V., and Burton, A. M. 2011. Variability in photos of the same face. Cognition 121, 313--323.Google ScholarCross Ref
- Johnston, A., Hill, H., and Carmen, N. 1992. Recognising faces: effects of lighting direction, inversion and brightness. Perception 21, 365--375.Google ScholarCross Ref
- Johnston, R. A. and Edmonds, A. J. 2009. Familiar and unfamiliar face recognition: A review. Memory 17, 5, 577--596.Google ScholarCross Ref
- Macmillan, N. A. and Creelman, C. D. 1991. Detection Theory: A User's Guide. Cambridge University Press, Cambridge, UK.Google Scholar
- Natu, V. and O'Toole, A. J. 2011. The neural processing of familiar and unfamiliar faces: A review and synopsis. British J. Psychol. 102, 726--747.Google ScholarCross Ref
- O'Toole, A., Abdi, H., Jiang, F., and Phillips, P. J. 2007. Fusing face recognition algorithms and humans. IEEE Trans. Syst. Man Cybern. 37, Part B, 1149--1155. Google ScholarDigital Library
- O'Toole, A., Roark, D., and Abdi, H. 2002. Recognition of moving faces: A psychological and neural perspective. Trends Cognitive Sci. 6, 261--266.Google ScholarCross Ref
- O'Toole, A. J., Phillips, P. J., An, X., and Dunlop, J. 2012. Demographic effects on estimates of automatic face recognition. Image, Vision, Comput. 30, 169--176. Google ScholarDigital Library
- O'Toole, A. J., Phillips, P. J., Jiang, F., Ayyad, J., Penard, N., and Abdi, H. 2007. Face recognition algorithms surpass humans matching faces across changes in illumination. IEEE Trans. Pattern Analy. Machine Intell. 29, 1642--1646. Google ScholarDigital Library
- O'Toole, A. J., Phillips, P. J., and Narvekar, A. 2008. Humans versus algorithms: Comparisons from the FRVT 2006. In Proceedings of the Eighth International Conference on Automatic Face and Gesture Recognition.Google Scholar
- Phillips, P. J., Beveridge, J. R., Draper, B. A., Givens, G., O'Toole, A. J., Bolme, D., Dunlop, J., Lui, Y. M., Sahizada, H., and Weimer, S. 2012. An introduction to the good, bad, and ugly challenge problem. Image. Vision, Comput. 30, 177--185. Google ScholarDigital Library
- Phillips, P. J., Beveridge, J. R., Draper, B. A., Givens, G., O'Toole, A. J., Bolme, D., Dunlop, J., Lui, Y. M., Sahibzada, H., and Weimer, S. 2011. An Introduction to the good, the bad, and the ugly face recognition challenge problem. In Proceedings of the 9th International Conference on Automatic Face and Gesture Recognition.Google Scholar
- Phillips, P. J., Flynn, P. J., Scruggs, W. T., Bowyer, K. W., Chang, J., Hoffmann, K., Marques, J., Min, J., and Worek, W. 2005. Overview face recognition grand challenge results. In Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR). I:947--954. Google ScholarDigital Library
- Phillips, P. J., Jiang, F., Narvekar, A., and O'Toole, A. J. 2010. An other-race e_ect for face recognition algorithms. ACM Trans. Appl. Percept. 8. Google ScholarDigital Library
- Phillips, P. J., Scruggs, W. T., O'Toole, A. J., Flynn, P. J., Bowyer, K. W., Schott, C. L., and Sharpe, M. 2010. FRVT 2006 and ICE 2006 large-scale results. IEEE Trans. Pattern Analy. Machine Intell. 32, 5, 831--846. Google ScholarDigital Library
- Rice, A., Phillips, P. J., Natu, V. S., An, X., and O'Toole, A. J. 2012. Unconscious use of the body in identifying the face. J. Vision, VSS Abstract.Google Scholar
Index Terms
- Comparing face recognition algorithms to humans on challenging tasks
Recommendations
Age-Invariant Face Recognition
One of the challenges in automatic face recognition is to achieve temporal invariance. In other words, the goal is to come up with a representation and matching scheme that is robust to changes due to facial aging. Facial aging is a complex process that ...
3D Face Factorisation for Face Recognition Using Pattern Recognition Algorithms
AbstractThe face is the preferable biometrics for person recognition or identification applications because person identifying by face is a human connate habit. In contrast to 2D face recognition, 3D face recognition is practically robust to illumination ...
Face recognition under varying illumination using gradientfaces
In this correspondence, we propose a novel method to extract illumination insensitive features for face recognition under varying lighting called the Gradientfaces. Theoretical analysis shows Gradientfaces is an illumination insensitive measure, and ...
Comments