Towards Facial Expression Analysis in a Driver Assistance System

摘要

We present first steps to enable facial expression recognition (FER) in the Smart Eye driver monitoring systems (DMS). To tailor the system to our specific active IR illumination, we record our own multi-camera expression database that is effectively annotated with bounding boxes around the face using the DMS. We compare different deep convolutional neural network architectures suitable for embedded devices on the classification task.