OEM & Lieferant Ausgabe 2/2019 - OEM & Supplier 2/2019 by VEK Publishing

133 at wavelengths under 780 nm,” says Wagner. “In thismanner wemade sure that the captured light came primarily from the IR LEDs and that their full functionality was assured during day and night time.” In addition, blocking visible daylight prevented shadow effects in the driver area that might otherwise have led to mistakes in facial recognition. A Raspberry Pi 3 Model B+ sent a trigger signal to both cameras to synchronize the moment of image capture. With this setup, images were captured of the postures of 16 test persons in a stationary ve- hicle. To generate a wide range of data, the test persons differed in gender, age, and head- gear, as well as using different mobile phone models and consuming different foods and beverages. “We set up five distraction cate- gories that driver postures could later be as- signed to. These were: ‘no visible distraction,’ ‘talking on smartphone,’ ‘manual smartphone use,’ ‘eating or drinking’ and ‘holding food or beverage,’” explains Wagner. “For the tests, we instructed the test persons to switch be- tween these activities during simulated driv- ing.” After capture, the images from the two cameras were categorized and used for model training. Training and testing the image classification systems Four modified CNN models were used to clas- sify driver postures: ResNeXt-34, ResNeXt-50, VGG-16 and VGG-19. The last two models are widely used in practice, while ResNeXt-34 and ResNeXt-50 contain a dedicated structure for processing of parallel paths. To train the system, ARRK ran 50 epochs using the Adam optimizer, an adaptive learning rate optimiza- tion algorithm. In each epoch, the CNN model had to assign the test persons’ postures to the defined categories. With each step, this categorization was adjusted by a gradient descent method, so that the fault rate could be lowered continuously. After model training, a dedicated test dataset was used to calculate the error matrix which allowed an analysis of the fault rate per driver posture for each CNN model. “The use of two cameras, each with a separately trained CNN model, enables ideal case differentiation for the left and right side of the face,” explains Wagner. “Thanks to this process, we were able to identify the system with the best performance in recognizing the use of mobile phones and consumption of food and beverages.” Evaluation of the results showed that the ResNeXt-34 and ResNeXt-50 models achieved the highest classification ac- curacy, 92.88 percent for the left camera and 90.36 percent for the right camera. Using this information, ARRK has extended its training database which now contains around 20,000 labeled eye data records. Based on this, it is possible to develop an automated vision-based system to validate driver mon- itoring systems. ARRK Engineering’s experts are already planning another step to further reduce the fault rate. Images: © ARRK Engineering ARRK Engineering www.arrk-engineering.com Website ARRK Engineering info@arrk-engineering.com eMail In the test setup, two cameras with active infrared lighting were positioned to the left and right of the driver on the A-column of a test vehicle. Both cameras had a frequency of 30 Hz and delivered 8-bit grayscale images at 1280 x 1024 pixel resolution. For the experiment, the test persons were instructed to switch among five different activities during simulated driving. After capture, the images from the two cameras were categorized and used for model training.