Hardware efficient FPGA implementation of emotion recognizer for autistic children

A Real-time emotion recognition system has potential applications especially for people suffering from autism to understand other people's emotions. A portable emotion recognizer will aid the autistic person to interact with the external world easily and helps them to understand facial emotions during their face to face communication. In this paper we have proposed a portable hardware efficient emotion recognizer using principal component analysis. The Eigen values are obtained using Jacobi iteration and the proposed architecture optimizes the Eigen calculation by using only diagonal and upper triangular matrix of the symmetric covariance matrix. The proposed emotion recognizer architecture is implemented on Virtex 7 XC7VX330T FFG1761-3 FPGA. We achieved 72.9 % detection accuracy for a word-length of 12 bit.