Nikolas Ioannou, Sage Khanuja
November 9, 2019
Drive.ai is a mobile application for the iPhone that implements a custom image classification algorithm to detect distracted driving and audibly alert drivers in real time. Drive.ai does not only detect distracted driving, but also provides the driver with valuable information as to how they were distracted, supporting 10 types of distracted driving.
We developed Drive.ai in Xcode using Swift. Drive.ai features a non-distracting screen while driving or a map with both visual and auditory directions. Furthermore, Drive.ai adjusts the color of its user interface to ensure that the phone is as least distracting as possible. Finally, Drive.ai includes a page that appears once the drive has culminated that displays valuable statistics about how the driver was primarily distracted to improve in the future.
We used Tensorflow with Keras and transfer learning on VGG19 to develop our image classification algorithm. After training, we received 93% validation accuracy.
CodeDay: The first iteration of Drive.ai was developed in 24 hours for the CodeDay Bellevue hackathon in 2019. Drive.ai received the Best in Show award.