Machine Learning for a Safer Tomorrow: Detecting Driver Distraction and Drowsiness
The rising number of accidents caused by distracted or drowsy drivers has spurred an interest in developing intelligent systems to detect these conditions and issue timely alerts to the driver. During this presentation, I will discuss about our collaboration with Australia’s foremost Telematics provider to create a computer vision-based solution for detecting distraction and drowsiness. Additionally, we’ll highlight the work done to establish a machine learning platform that enables their team to iterate over the solution and improve the results.
Watch the Brownbag
Meet the host
- Senior Data Scientist
- Auckland, New Zealand
Senior Data Scientist
Driver distraction is typically divided into three categories: physical, visual, and cognitive. Physical distractions include things like eating, drinking, or using a mobile device while driving. Visual distractions take a driver’s eyes off the road, such as looking at a map or checking a phone. Cognitive distractions are those that take a driver’s mind off the task of driving, like daydreaming or being emotionally upset. For the scope of the project, we focused on detecting physical distractions. Similarly for drowsiness, the focus was to detect microsleep which is a brief and involuntary episode of sleep that can last for a few seconds to several minutes.
In this brown bag session, we will start the presentation by going through the background of the project and why one should care about distraction and drowsiness. Next, we will go through the different steps involved in implementing a machine learning solution and how these were applied for the distraction and drowsiness detection solution. We will then go through the model and application evaluation results. Lastly, we will look at some videos showcasing the detection of distraction and drowsiness events.