Evaluation of facial analysis methods for personalization in driver monitoring systems

Pham T. A., Kashevnik M. V., Chechulin A. A.

Recently, research and development on the topic of driver monitoring systems have become more and more popular. Since self-driving cars remain the long perspective, the research & development community in the field of intelligent transportation concentrated on driver monitoring in-vehicle cabin to reduce the number of traffic accidents. The paper presents an analysis of the relevant work and reviews modern methods for identifying potentially distracting driver situations based on the analysis of images received from the front camera of the smartphone. The study aims to answer the question: how to set up methods for analyzing facial images for different types of faces (for example, in Europe or Asia) to personalize the system and increase the accuracy of facial features detection. Progress in the field of deep learning is widely used in modern identification systems and can also be applied to the problem under consideration. The paper assesses the advantages and disadvantages of the existing machine learning methods in relation to the problem under consideration.

Keywords:Face analysis, personification, deep learning, artificial neural networks, convolutional neural networks