Gender and ethnicity classification using deep learning in heterogeneous face recognition

Although automated classification of soft biometric traits in terms of gender, ethnicity and age is a well-studied problem with a history of more than three decades, it is still far from being considered a solved problem for the case of difficult exposure conditions, such as during night-time, in environments with unconstrained lighting, or at large distances from the camera. In this paper, we investigate the advantages and limitations of the automated classification of soft biometric traits in terms of gender and ethnicity in Near-Infrared (NIR) long-range, night-time face images. The impact of soft biometric traits in terms of gender and ethnicity is explored for the purpose of improving cross-spectral face recognition (FR) performance. The main contributions are, (i) a dual database collected in NIR band at night time and at four different distances of 30, 60, 90 and 120 meters is used, (ii) a deep convolution neural network to perform the classification in terms of gender and ethnicity is proposed, (iii) a set of experiments is performed indicating that, the usage of soft biometric traits to perform face matching, resulted in a significantly improved rank-1 identification rate when compared to the original biometric system (scenario dependent).