PROJECT TYPE:

• Research Project

PROJECT DURATION:

• 1.10.2025 – 30.09.2029.

PROJECT LEADER AND ASSOCIATES:

• Assoc. Prof. Ph.D. Goran Oreški (Leader),
• Prof. Ph.D. Tihomir Orehovački,
• Prof. Ph.D. Antoni Granollers Saltiveri,
• Assoc. Prof. Ph.D. Dijana Oreški,
• Assoc. Prof. Ph.D. Alicia García Holgado,
• Asst. Prof. Ph.D. Nikola Tanković,
• Ph.D. Robert Šajina,
• Romeo Šajina, M.Inf.

BRIEF PROJECT DESCRIPTION:

The DMM (Driver Monitoring Model) project – Development of a Model for Monitoring Driver Attention and Fatigue Using Artificial Intelligence – is focused on addressing a critical road safety issue, as driver errors, such as inattention and fatigue, are the cause of almost all traffic accidents (96.1% with an established cause) in the Republic of Croatia. The main goal of the project is to develop an advanced artificial intelligence algorithm based on deep learning and computer vision, aiming for reliable and timely recognition of signs of fatigue and reduced attention in drivers in real-time. The project methodology focuses on precise monitoring of key biometric parameters of the driver's face. These parameters include analyzing eye movements, blink frequency, measuring eyelid closure (PERCLOS), and changes in head position and orientation, gaze direction, and facial expressions. The key innovation lies in developing a solution that will be non-invasive and robust, focusing on the efficiency of using a single camera (mono-RGB and/or near-infrared) under various driving conditions. Additionally, an optimal user interface will be defined so that warnings are effective while minimally distracting the driver.

PROJECT OBJECTIVES:

The main strategic objective of the DMM project is to essentially improve road safety by directly reducing the number of traffic accidents caused by human errors, particularly driver inattention and fatigue. Since driver errors are the main cause of accidents, the primary goal is to develop a proactive, technological solution that automatically recognizes and prevents the occurrence of critical situations on the road.

The specific objectives of the project include:

• Development of an advanced driver state detection algorithm: Creating a robust algorithm based on deep learning and computer vision, capable of reliably recognizing early signs of fatigue and reduced attention in real-time.
• Defining key biometric parameters: Determining and implementing a set of parameters – such as the PERCLOS metric (eyelid closure ratio), blink frequency, gaze direction, head position and orientation – that accurately correlate with the driver's fatigue level.
• Ensuring system robustness and non-invasiveness: The goal is to develop a model that effectively functions across a wide range of real driving conditions (varying lighting, ethnic differences, use of glasses) using a minimal, non-invasive camera configuration (mono-RGB and/or near-infrared).
• Designing an optimal and non-distracting user interface: Development and evaluation of an effective warning system that minimally distracts the driver from the road, while simultaneously ensuring a quick reaction to detected fatigue or inattention.

PROJECT HOLDER (INSTITUTION): Juraj Dobrila University of Pula