Lane violation detection using convolutional neural networks
Disclosed herein are systems, methods, and devices for detecting traffic lane violations. In one embodiment, a method for detecting a potential traffic violation is disclosed comprising bounding a vehicle detected from one or more video frames of a video in a vehicle bounding box. The vehicle can be detected and bounded using a first convolutional neural network. The method can also comprise bounding, using the one or more processors of the edge device, a plurality of lanes of a roadway detected from the one or more video frames in a plurality of polygons. The plurality of lanes can be detected and bounded using multiple heads of a multi-headed second convolutional neural network. The method can further comprise detecting a potential traffic violation based in part on an overlap of at least part of the vehicle bounding box and at least part of one of the polygons.
1. A method for detecting a potential traffic violation, comprising:
bounding, using one or more processors of an edge device, a vehicle detected from one or more video frames of one or more videos in a vehicle bounding polygon;
bounding, using the one or more processors of the edge device, a lane-of-interest of a roadway detected from the one or more video frames in a LOI polygon;
generating a three-dimensional (3D) bounding polygon based on the vehicle bounding polygon, wherein a polygonal base of the 3D bounding polygon represents a road surface underneath the vehicle;
determining a lane occupancy score by computing a degree of overlap of the polygonal base with the LOI polygon; and
detecting, using one or more processors of an edge device or a server, a potential traffic violation when the lane occupancy score exceeds a predetermined threshold value.
2. The method of claim 1 , wherein the polygonal base is shaped as a quadrilateral.
3. The method of claim 1 , wherein the one or more videos are captured by one or more video image sensors of the edge device.
4. The method of claim 1 , wherein the vehicle is detected and bounded using a first machine learning algorithm.
5. The method of claim 4 , wherein the LOI polygon is detected and bounded using a second machine learning algorithm separate from the first machine learning algorithm.
6. The method of claim 1 , wherein the edge device is coupled to a carrier vehicle and wherein the one or more videos are captured by the edge device while the carrier vehicle is in motion.
7. A device for detecting a potential traffic violation, comprising:
one or more video image sensors configured to capture one or more videos of a vehicle and a plurality of lanes of a roadway; and
one or more processors programmed to execute instructions to:
bound the vehicle detected from one or more video frames of the one or more videos in a vehicle bounding polygon;
bound a lane-of-interest of a roadway detected from the one or more video frames in a LOI polygon;
generate a three-dimensional ( 3 D) bounding polygon based on the vehicle bounding polygon, wherein a polygonal base of the 3 D bounding polygon represents a road surface underneath the vehicle;
determine a lane occupancy score by computing a degree of overlap of the polygonal base with the LOI polygon; and
detect a potential traffic violation when the lane occupancy score exceeds a predetermined threshold value.
8. The device of claim 7 , wherein the polygonal base is shaped as a quadrilateral.
9. The device of claim 7 , wherein the vehicle is detected and bounded using a first machine learning algorithm.
10. The device of claim 9 , wherein the LOI polygon is detected and bounded using a second machine learning algorithm separate from the first machine learning algorithm.
11. The device of claim 7 , wherein the device is coupled to a carrier vehicle and wherein the one or more videos are captured by the device while the carrier vehicle is in motion.
12. One or more non-transitory computer-readable media comprising instructions stored thereon, that when executed by one or more processors, perform steps comprising:
bounding a vehicle detected from one or more video frames of one or more videos in a vehicle bounding polygon;
bounding a lane-of-interest of a roadway detected from the one or more video frames in a LOI polygon;
generating a three-dimensional (3D) bounding polygon based on the vehicle bounding polygon, wherein a polygonal base of the 3D bounding polygon represents a road surface underneath the vehicle;
determining a lane occupancy score by computing a degree of overlap of the polygonal base with the LOI polygon; and
detecting a potential traffic violation when the lane occupancy score exceeds a predetermined threshold value.
13. The one or more non-transitory computer-readable media of claim 12 , wherein the polygonal base is shaped as a quadrilateral.
14. The one or more non-transitory computer-readable media of claim 12 , wherein the one or more videos are captured by one or more video image sensors of an edge device.
15. The one or more non-transitory computer-readable media of claim 14 , wherein the edge device is coupled to a carrier vehicle.
16. The one or more non-transitory computer-readable media of claim 15 , wherein the one or more videos are captured by the edge device while the carrier vehicle is in motion.
17. The one or more non-transitory computer-readable media of claim 12 , wherein the vehicle is detected and bounded using a first machine learning algorithm.
18. The one or more non-transitory computer-readable media of claim 17 , wherein the LOI polygon is detected and bounded using a second machine learning algorithm separate from the first machine learning algorithm.
19. A method for detecting a potential traffic violation, comprising:
bounding, using one or more processors of an edge device, a vehicle detected from one or more video frames of one or more videos in a vehicle bounding polygon, wherein the one or more videos are captured by one or more video image sensors of the edge device, and wherein the vehicle is detected and bounded using a first machine learning algorithm;
bounding, using the one or more processors of the edge device, one or more lanes of a roadway detected from the one or more video frames in one or more polygons, wherein the one or more lanes are detected and bounded using a second machine learning algorithm separate from the first machine learning algorithm, and wherein at least one of the polygons is a lane-of-interest (LOI) polygon bounding a LOI; and
detecting, using the one or more processors, a potential traffic violation based in part on an overlap of at least part of the vehicle bounding polygon and at least part of the LOI polygon.
20. The method of claim 19 , wherein the edge device is coupled to a carrier vehicle and wherein the one or more videos are captured by the edge device while the carrier vehicle is in motion.