Automatic collection and classification of harsh driving events in dashcam videos

1. A method, comprising:

receiving, by a device, a first message,

wherein the first message includes a link to video data and information concerning a vehicle and a driving event,

wherein the video data includes a plurality of video frames captured by one or more cameras associated with the vehicle;

obtaining, by the device, the video data using the link;

identifying, by the device, the vehicle and the driving event based on the information concerning the vehicle and the driving event;

determining, by the device, one or more objects in a video frame of the plurality of video frames;

constructing, by the device, a collision cone of the vehicle for the video frame;

determining, by the device and based on the collision cone of the vehicle, a set of objects of the one or more objects in the video frame that have a potential to collide with the vehicle;

assigning, by the device, a category to the driving event based on the set of objects, the collision cone of the vehicle, and the information concerning the vehicle and the driving event;

generating, by the device, a second message that includes the category and the information concerning the vehicle and the driving event; and

sending, by the device, the second message to a client device to permit the client device to display the second message.

2. The method of claim 1 , wherein determining the one or more objects in the video frame of the plurality of video frames comprises:

processing the video frame using an object detection algorithm,

wherein the object detection algorithm employs a convolutional neural network.

3. The method of claim 1 , wherein constructing the collision cone of the vehicle for the video frame comprises:

computing an optical flow of the video frame;

determining a vanishing point of the video frame; and

constructing the collision cone of the vehicle for the video frame based on the optical flow and the vanishing point.

4. The method of claim 1 , wherein determining, based on the collision cone of the vehicle, the set of objects of the one or more objects in the video frame that have the potential to collide with the vehicle comprises:

determining motion of the one or more objects;

calculating a time to collision of the one or more objects based on the collision cone; and

determining the set of objects based on the motion of the one or more objects and the time to collision of the one or more objects.

5. The method of claim 1 , wherein assigning the category to the driving event based on the set of objects, the collision cone of the vehicle, and the information concerning the vehicle and the driving event comprises:

processing information regarding the set of objects, information regarding the collision cone of the vehicle, and the information concerning the vehicle and the driving event using a vehicle event categorization algorithm,

wherein the vehicle event categorization algorithm is a machine learning algorithm.

6. The method of claim 1 , wherein the category indicates a level of severity of the driving event.

7. The method of claim 6 , wherein the category is based on:

a duration of the driving event;

an acceleration or deceleration rate of the vehicle;

a speed of the vehicle; and

a trigger of the driving event.

8. A device, comprising:

one or more memory devices; and

one or more processors operatively coupled to the memory devices, the one or more processors to:

receive a first message,

wherein the first message includes a link to video data and information concerning a vehicle and a driving event,

wherein the video data includes a plurality of video frames captured by one or more cameras associated with the vehicle;

obtain the video data using the link;

identify the vehicle and the driving event based on the information concerning the vehicle and the driving event;

determine one or more objects in a video frame of the plurality of video frames;

construct a collision cone of the vehicle for the video frame;

determine, based on the collision cone of the vehicle, a set of objects of the one or more objects in the video frame that have a potential to collide with the vehicle;

assign a category to the driving event based on the set of objects, the collision cone of the vehicle, and the information concerning the vehicle and the driving event;

generate a second message that includes the category and the information concerning the vehicle and the driving event; and

send the second message to a client device to permit the client device to display the second message.

9. The device of claim 8 , wherein the one or more processors, when determining the one or more objects in the video frame of the plurality of video frames, are to:

process the video frame using an object detection algorithm,

wherein the object detection algorithm employs a convolutional neural network.

10. The device of claim 8 , wherein the one or more processors, when constructing the collision cone of the vehicle for the video frame, are to:

compute an optical flow of the video frame;

determine a vanishing point of the video frame; and

construct the collision cone of the vehicle for the video frame based on the optical flow and the vanishing point.

11. The device of claim 8 , wherein the one or more processors, when determining, based on the collision cone of the vehicle, the set of objects of the one or more objects in the video frame that have the potential to collide with the vehicle, are to:

determine motion of the one or more objects;

calculate a time to collision of the one or more objects based on the collision cone; and

determine the set of objects based on the motion of the one or more objects and the time to collision of the one or more objects.

12. The device of claim 8 , wherein the one or more processors, when assigning the category to the driving event based on the set of objects, the collision cone of the vehicle, and the information concerning the vehicle and the driving event, are to:

process information regarding the set of objects, information regarding the collision cone of the vehicle, and the information concerning the vehicle and the driving event using a vehicle event categorization algorithm,

wherein the vehicle event categorization algorithm is a machine learning algorithm.

13. The device of claim 8 , wherein the category indicates a level of severity of the driving event.

14. The device of claim 13 , wherein the category is based on:

a duration of the driving event;

an acceleration or deceleration rate of the vehicle;

a speed of the vehicle; and

a trigger of the driving event.

15. A non-transitory computer-readable medium storing instructions, the instructions comprising:

one or more instructions that, when executed by one or more processors, cause the one or more processors to:

receive, by a device, a first message,

wherein the first message includes a link to video data and information concerning a vehicle and a driving event,

wherein the video data includes a plurality of video frames captured by one or more cameras associated with the vehicle;

obtain the video data using the link;

identify the vehicle and the driving event based on the information concerning the vehicle and the driving event;

determine one or more objects in a video frame of the plurality of video frames;

construct a collision cone of the vehicle for the video frame;

determine, based on the collision cone of the vehicle, a set of objects of the one or more objects in the video frame that have a potential to collide with the vehicle;

assign a category to the driving event based on the set of objects, the collision cone of the vehicle, and the information concerning the vehicle and the driving event;

generate a second message that includes the category and the information concerning the vehicle and the driving event; and

send the second message to a client device to permit the client device to display the second message.

16. The non-transitory computer-readable medium of claim 15 , wherein the one or more instructions, that cause the one or more processors to determine the one or more objects in the video frame of the plurality of video frames, cause the one or more processors to:

process the video frame using an object detection algorithm,

wherein the object detection algorithm employs a convolutional neural network.

17. The non-transitory computer-readable medium of claim 15 , wherein the one or more instructions, that cause the one or more processors to construct the collision cone of the vehicle for the video frame, cause the one or more processors to:

compute an optical flow of the video frame;

determine a vanishing point of the video frame; and

construct the collision cone of the vehicle for the video frame based on the optical flow and the vanishing point.

18. The non-transitory computer-readable medium of claim 15 , wherein the one or more instructions, that cause the one or more processors to determine, based on the collision cone of the vehicle, the set of objects of the one or more objects in the video frame that have the potential to collide with the vehicle, cause the one or more processors to:

determine motion of the one or more objects;

calculate a time to collision of the one or more objects based on the collision cone; and

determine the set of objects based on the motion of the one or more objects and the time to collision of the one or more objects.

19. The non-transitory computer-readable medium of claim 15 , wherein the one or more instructions, that cause the one or more processors to assign the category to the driving event based on the set of objects, the collision cone of the vehicle, and the information concerning the vehicle and the driving event, cause the one or more processors to:

process information regarding the set of objects, information regarding the collision cone of the vehicle, and the information concerning the vehicle and the driving event using a vehicle event categorization algorithm,

wherein the vehicle event categorization algorithm is a machine learning algorithm.

20. The non-transitory computer-readable medium of claim 15 , wherein the category is based on:

a duration of the driving event;

an acceleration or deceleration rate of the vehicle;

a speed of the vehicle; and

a trigger of the driving event.