Staying Awake, Staying Alive with an AI-Powered Dashcam
Imagine driving down an empty highway late at night. The road stretches endlessly, your eyelids grow heavier, and in the next moment…a gentle beep nudges you back to alertness. And no, that beep isn’t just a sound. It’s an AI-powered dashcam designed to keep you safe, a project brought to life by a group of freshmen at CMKL who saw a problem too big to ignore: drowsy driving.
It all started with some alarming statistics. Drowsy driving causes nearly 20% of all car crashes in the U.S. alone. The team realized that the usual solutions—awareness campaigns, driver self-assessments—weren’t enough. They needed something that didn’t rely on people realizing they were tired. Thus, they built a dashcam that does the noticing for them. It records, analyzes, and alerts, using AI to detect signs of fatigue like heavy eyelids and head nodding before it's too late.
The foundation of the project is built upon advanced technologies, primarily using YOLOv8, a real-time object detection algorithm. The team also used Apex, a supercomputer hosted by CMKL, to train custom datasets to improve the model's accuracy in identifying signs of driver fatigue, including prolonged eye closure, head nodding, and changes in facial expressions. Additionally, they designed an intuitive app interface using Figma, because what’s the point of great tech if it’s not easy to use?
Of course, the journey was anything but smooth. The team faced significant obstacles in data training, particularly concerning model accuracy under varying lighting conditions and the presence of obstructions such as sunglasses and face masks.
"If I could go back, I’d start sooner," one of them admitted, reflecting on how much time was eaten up just troubleshooting. But their professor disagreed. "I wouldn’t change a thing," he said. "They learned the most from the hard parts. That’s the point."
Privacy concerns were an important consideration throughout the project. The team implemented multi-layered measures to safeguard user data, ensuring that all footage is stored locally on the device with no automatic cloud uploads. "All the data stays on the device," they explained. "Nothing goes to the cloud unless the user says so." Simple, transparent, and in control of the person who matters most—the driver.
The outcomes of the project are remarkable. The dashcam effectively detects signs of drowsiness and provides timely alerts, significantly enhancing driver safety. Moving forward, they plan to refine the system further, incorporating features like night mode and advanced alert mechanisms, thereby contributing to broader road safety initiatives.
Their dashcam works.
It detects drowsiness.
It keeps people awake and alive.
And what matters most is that they’re using technology where it counts the most: to save lives. Who knows how many lives will be saved in the future because of this? And that, in itself, is extraordinary.
