Sparse data representation is still a very new concept (developed as a mathematical model of human long term memory by Pentti Kanerva in 1992. (Kanerva, 1992) While it shows great promise in tasks like face recognition, there is no open source library that puts this technology into public hands. By creating such a library, this will help advance the use and understanding of sparse data representation by exposing the technology to those who would not otherwise be able to use it.

Facial recognition is especially important because it is one of the most basic functions of the human brain and one of the few things we are actually born pre-programmed to do. (McKone et al, 2009) Since much of human communication relies on body language, and much of that is concentrated in the face, being able to follow and recognize faces is one of the most basic tasks inherent in naturalistic human communication. (Schmidt and Cohn, 2001)
Also, by implementing a function similar to one of the basic functionalities of the human brain, in a way inspired by the way our brains process data, we are taking the first steps toward an undeniably conscious created silicone lifeform. This brings with it many fascinating ethical and practical implications. By keeping this technology open source, it facilitates open public discourse on the subject.

To this end, I am working on a project creating an open source face recognition library. I am creating this library using Walnut AI (an open source java implementation of neural network architecture). (Liu) I will be employing OpenCV (OpenCV), an open source image processing library, to preprocess the input to a form more conducive to recognition (eliminating extraneous information and minimizing variance in pose, expression and lighting). This system will implement the algorithm described in “Sparse Representation for Computer Vision and Pattern Recognition” (Wright et al, 2010), and will make use of training data provided on matlab. (Sri Hari, 2011) I will be demonstrating the finished library on the Atlas platform and I will be testing its effectiveness by comparison with project oxford (Project Oxford), a commercially available face recognition software as a service API.

References:

Kanerva, Pentti. “Sparse distributed memory and related models.” (1992).McKone, Elinor, Kate Crookes, and Nancy Kanwisher. “The cognitive and neural development of face recognition in humans.” The cognitive neurosciences 4 (2009): 467-482.

Schmidt, Karen L., and Jeffrey F. Cohn. “Human facial expressions as adaptations: Evolutionary questions in facial expression research.” American journal of physical anthropology 116.S33 (2001): 3-24.

Liu, Q. Walnut AI. Computer software. Web.
OpenCV. Computer software. Web.

Wright, John, et al. “Sparse representation for computer vision and pattern recognition.” Proceedings of the IEEE 98.6 (2010): 1031-1044.

Sri Hari, B. H. Robust Face Recognition via Sparse Representation – Implementation. Computer software. MathWorks File Exchange, 30 Mar. 2011. Web. 9 Mar. 2016.

Project Oxford. Computer software. Microsoft. Web.