Using Real-time Facial Recognition for Vehicle Driver Authentication

点击翻译按钮获取中文摘要

Smart vehicles are rapidly becoming a reality with advances in sensor technologies, machine learning, processing power, and internet access. With advances and popularization of intelligent devices and environments such as vehicles, it is increasingly critical that security and authentication be robust and reliable. Facial recognition for authentication of drivers is a real possibility that needs to be further studied, fine-tuned, and tested. In this project, we aim to tackle a number of important research questions associated with facial recognition for driver authentication in smart vehicles. These questions include i) identifying the minimum number, orientation, and setup of cameras required to reliably and robustly perform the task at hand, ii) developing robust models capable of recognizing and authenticating human faces with minimal training samples, under challenging conditions, and with changes over time, iii) identifying the optimum time for re-identification (short-term) as well as re-training (long-term), and iv) how users feel about and interact with this kind of technology. The project will train a postdoctoral fellow, a PhD student, and two master's students. The main beneficiary of this work will be Irdeto Canada, the Canadian arm of an international software and security company. The company will take immediate steps towards productization and commercialization of the technology with current customers having already shown interest. Additionally, in recent years, Canada has emerged as a global leader in artificial intelligence and machine learning, which has already resulted in significant economic benefits and growth. This project will contribute to Canada's AI ecosystem through development of novel algorithms and intelligent systems in the fields of image processing and security.