Overview / Usage

Distracted driving is a well-established problem, so much so that many states currently have bans in place when it comes to using technology while driving. But the problem of distracted walking is a relatively new one. Each year, more and more people are injured as a result of texting, talking or listening to music while on their cell phones. According to data from the National Highway Traffic Safety Administration (NHTSA), nearly 5,000 pedestrians were killed and an estimated 76,000 injured in traffic collisions in the United States in 2012. That’s one death every 2 hours and an injury every 7 minutes

Methodology / Approach

To solve this problem I used a modified Android launcher, a Raspberry Pi, the rear camera and the Intel Movidius stick to determine is the user is about to cross the street without paying attention to traffic.

Once the Android launcher is instantiated, it automatically takes pictures every 3 seconds using the rear camera without user interaction. The images are then forwarded to the Intel Movidius Tensorflow Model, then they are classified into a set of pre-trained categories. If the Tensorflow model finds a crosswalk or the edge of the sidewalk, then the phone is blocked and a pop up is displayed warning the user to look up.

Technologies Used

Android, Raspberry Pi, Intel Movidius and Intel Powered PC