Acceso usuarios
San Miguel De Ibarra, Ecuador
Traffic accidents due to drowsiness and distraction are significant problems in Ecuador. In the present study, a prototype of an Android mobile application was developed, for the detection and alertness of sleepiness in night driving, in real time, through the use of artificial vision techniques. This project was carried out through the Android Studio Development IDE, OpenCV library, Haar’Like Cascade Classifiers and Template matching techniques under the XP development methodology. The mobile application detects the location and dimensions of a face, within the image acquired in the visible spectrum. Then, these values are used to obtain the region of interest that contains each eye, using geometric equivalences applied to the human face. Subsequently, the presence of pupils in the region of interest is detected, applying template matching techniques and the fast Fourier transform, to determine whether the driver’s eyes are open or not. Finally, an audible alarm is issued if pupils are not detected in the ROI for a determined period of time. After a testing and configuring process, the developed application achieved an average detection rate of 32 fps, with a drowsiness detection accuracy of 91.46% and an AUC of 0.847.
