Introduction

The facial authentication system is based exclusively on depth images acquired by infrared sensors, in order to guarantee correct operation under different lighting conditions. The portability of this invention is guaranteed by the use of an embedded board equipped with a graphic processing unit.

Technical features

We present an authentication system for low-light-level environments. Input images are then sent to an embedded platform (like Nvidia Jetson), equipped with a Graphics Processing Unit (GPU), able to provide after a few tens of milliseconds the output of the system, i.e. a continuous value indicating the belonging of the subject to one of the identities previously registered in the system. The use of images from depth sensors makes it possible to obtain a functioning system even in the absence or partial presence of light sources, as well as to satisfy constraints regarding the maintenance of user privacy.

Possible Applications

• Recognition of the identity of the driver of a vehicle, even at night or in conditions of limited visibility.

Advantages

• Working in low-light-level environments;
• High level performance;
• Real Time processing;
• Low hardware and energy consumption requirements.