Driver Drowsiness Detection by Applying Deep Learning Techniques to Sequences of Images

This work presents the development of an ADAS (advanced driving assistance system) focused on driver drowsiness detection, whose objective is to alert drivers their drowsy state avoid road traffic accidents. In a environment, it necessary that fatigue detection performed in non-intrusive way, and not bothered with alarms when he or she drowsy. Our approach this open problem uses sequences images are 60 s long recorded such way subject’s face visible. To detect whether shows symptoms not, two alternative solutions developed, focusing minimization false positives. The first recurrent convolutional neural network, while second one deep learning techniques extract numeric features from images, which introduced into fuzzy logic-based system afterwards. accuracy obtained by both systems similar: around 65% over training data, 60% test data. However, stands out because avoids raising reaches specificity (proportion videos correctly classified) 93%. Although results do achieve very satisfactory rates, proposals presented promising can be considered solid baseline for future works.