| CPC G08B 21/06 (2013.01) [B60W 30/0956 (2013.01); B60W 40/08 (2013.01); B60W 40/10 (2013.01); B60W 50/14 (2013.01); G06F 18/21 (2023.01); G06N 3/04 (2013.01); G06N 3/08 (2013.01); G06V 20/40 (2022.01); G06V 20/58 (2022.01); G06V 20/597 (2022.01); G08G 1/20 (2013.01); H04N 7/183 (2013.01); H04N 23/54 (2023.01); B60W 2420/403 (2013.01); B60W 2510/18 (2013.01); B60W 2510/20 (2013.01); B60W 2520/00 (2013.01); B60W 2540/225 (2020.02); B60W 2555/20 (2020.02); G06V 20/44 (2022.01)] | 13 Claims |
|
1. A driver monitoring system, comprising:
a gaze tracking camera configured to record images of a driver within a cabin of a vehicle and determine a gaze direction of the driver in the recorded images; and
a controller in communication with the gaze tracking camera, operatively connected to memory, and configured to:
detect a potential distracted driving event based on the gaze direction of the driver as depicted in a particular image of the recorded images being outside of a predefined alert driver area for an amount of time exceeding a predefined time threshold;
provide the particular image to a convolutional neural network that has been trained with images depicting drivers utilizing mobile devices;
determine, based on feedback from the convolutional neural network, whether the driver is utilizing a mobile device in the particular image;
based on the determination indicating that the driver is utilizing the mobile phone in the particular image, perform one or both of the following predefined actions: transmit the particular image to a fleet manager, and store the particular image in a local repository of anomalous driving images in the memory;
based on the determination indicating that the driver is not utilizing the mobile phone in the particular image, omit performance of said one or both of the predefined actions, and further train the convolutional neural network using the particular image;
adjust one or both of the predefined time threshold and the predefined alert driver area based on at least one of traffic density, weather conditions, object detection external to the vehicle, and a geographic location of the vehicle; AND
select one or both of the predefined time threshold and the predefined alert driver area based on an experience level of the driver.
|