What is IFS-DriverAlert

IFS-DriverAlert is a safety system designed to monitor a driver's face in real time and detect signs of drowsiness or distraction, such as eye closure, yawning, and looking away. It uses a camera, a Raspberry Pi 5, and MediaPipe AI models to analyze facial features and trigger alerts through a speaker. The goal is to reduce accidents caused by fatigue and inattention, especially in regular vehicles that don't have built-in driver monitoring systems.

The system has an Auto and a Manual modes. In Auto mode, the detection system activates only when the car speed is at least 20km/h, which is done using a NEO-6M GPS module to monitor the speed. This prevents unnecessary alerts when the car is stopped. In Manual mode, detection runs continuously regardless of speed, which is useful in cases like tunnels where GPS signals may be weak. An audio reminder also plays every 10 minutes if Auto mode is on but speed is not detected, recommending the driver to switch to Manual mode.

Why the System Matters

In Saskatchewan, Canada, driver fatigue causes over 145 injuries and 7 deaths every year on average, according to SGI. Drowsy driving is treated as seriously as impaired driving due to its impact on reaction time and judgment.

In the United States, the CDC reports that up to 7,500 fatal crashes annually (about 25% of all fatal crashes) involve drowsy driving. Nearly 1 in 25 adults admitted to falling asleep while driving in the past month.

These numbers show the urgent need for an affordable and effective drowsiness detection system. Although drowsiness detection systems are available in high-end and premium vehicles, they are often too expensive for most drivers. IFS-DriverAlert offers a low-cost, offline alternative that brings this life-saving technology to any standard vehicle.

Implementation Approach

A camera captures real-time video of the driver's face.
MediaPipe and OpenCV are used to detect eye closure, yawning, and looking away.
When signs of drowsiness or distraction are detected, a progressive audio alert is played through a speaker.
The system is powered by the car's 12V outlet using a step-down converter to provide 5V to the Raspberry Pi 5.
A NEO-6M GPS module monitors vehicle speed to enable Auto mode.
In Auto mode, detection only runs when the car speed is at least 20km/h, reducing power and avoiding false alerts when parked.
In Manual (On) mode, detection runs continuously, useful in areas like tunnels where GPS may not work.
An audio reminder plays every 10 minutes if Auto mode is active but speed is not detected, prompting the driver to manually turn detection on.
The system operates entirely offline, with no data storage or transmission, ensuring privacy and low power usage.

Our Solution

IFS-DriverAlert provides an affordable, standalone, and offline drowsiness detection system that can be added to any standard vehicle. It uses AI-powered facial monitoring to detect signs of fatigue and responds with progressive audio alerts to help keep drivers safe. This system ensures privacy by processing data locally without storing any images or video.

The below video shows the system components and operation.