We've already seen numerous systems that can detect if drivers are getting drowsy. Now, however, engineering researchers at Canada's University of Waterloo have created software that can tell if the driver is engaging in distracting activities such as texting.
Developed by a team led by Prof. Fakhri Karray at the university's Centre for Pattern Analysis and Machine Intelligence (CPAMI), the system uses in-car cameras and computer algorithms to watch for hand movements that deviate from those normally associated with driving. Those algorithms, which were "trained" using machine-learning techniques, are able to match the movements to known activities like texting, having a cell phone conversation, or reaching over into the back seat.
Once identified, distracting activities are graded according to the safety threat that they pose – this is based not only on what the activity is, but also on factors such as its duration. In cases where the distraction level is deemed to be too high, the system could warn the driver, or in extreme cases even temporarily take over the driving of the car.
Karray and colleagues are now looking at integrating other distracted-driving indicators into the system, including the monitoring of head and face position.
They recently presented their research at the 14th International Conference on Image Analysis and Recognition in Montreal.
Source: University of Waterloo
We've already seen numerous systems that can detect if drivers are getting drowsy. Now, however, engineering researchers at Canada's University of Waterloo have created software that can tell if the driver is engaging in distracting activities such as texting.
Developed by a team led by Prof. Fakhri Karray at the university's Centre for Pattern Analysis and Machine Intelligence (CPAMI), the system uses in-car cameras and computer algorithms to watch for hand movements that deviate from those normally associated with driving. Those algorithms, which were "trained" using machine-learning techniques, are able to match the movements to known activities like texting, having a cell phone conversation, or reaching over into the back seat.
Once identified, distracting activities are graded according to the safety threat that they pose – this is based not only on what the activity is, but also on factors such as its duration. In cases where the distraction level is deemed to be too high, the system could warn the driver, or in extreme cases even temporarily take over the driving of the car.
Karray and colleagues are now looking at integrating other distracted-driving indicators into the system, including the monitoring of head and face position.
They recently presented their research at the 14th International Conference on Image Analysis and Recognition in Montreal.
Source: University of Waterloo
