DEVICES AND METHODS FOR DETERMINING REGION OF INTEREST FOR OBJECT DETECTION IN CAMERA IMAGES

1 . An apparatus comprising:

a first camera configured to view an environment outside a vehicle; and

a processing unit configured to receive a first image from the first camera;

wherein the processing unit is configured to determine a centerline of a road or a lane in which the vehicle is traveling; and

wherein the processing unit is configured to determine a region of interest based on the centerline, the region of interest having a geometry that is based on a feature of the centerline.

2 . The apparatus of claim 1 , wherein the processing unit is configured to determine the centerline of the road or the lane in which the vehicle is traveling using on a neural network model.

3 . The apparatus of claim 1 , wherein the processing unit is configured to determine the centerline of the road or the lane in which the vehicle is traveling by determining a plurality of centerline points.

4 . The apparatus of claim 3 , wherein the centerline points comprise a first centerline point and a second centerline point, and wherein the processing unit is configured to determine the region of interest based on the first centerline point and the second centerline point.

5 . The apparatus of claim 4 , wherein the processing unit is configured to determine the region of interest by:

determining a first left point that is a distance d1 to a left of the first centerline point;

determining a first right point that is the distance d1 to a right of the first centerline point;

determining a second left point that is a distance d2 to a left of the second centerline point; and

determining a second right point that is the distance d2 to a right of the second centerline point.

6 . The apparatus of claim 5 , wherein the processing unit is configured to determine the region of interest also by:

determining a left boundary of the region of interest based on the first left point and the second left point; and

determining a right boundary of the region of interest based on the first right point and the second right point.

7 . The apparatus of claim 3 , wherein the centerline points have at least a first centerline point with a first y-coordinate, a second centerline point with a second y-coordinate, and a third centerline point with a third y-coordinate; and

wherein a first difference between the first y-coordinate and the second y-coordinate is different from a second difference between the second y-coordinate and the third y-coordinate.

8 . The apparatus of claim 1 , wherein the processing unit is configured to determine the region of interest by: determining a left boundary of the region of interest based on the determined centerline, and determining a right boundary of the region of interest based on the determined centerline.

9 . The apparatus of claim 8 , wherein a first distance between the centerline and the left boundary of the region of interest is non-constant, and a second distance between the centerline and the right boundary of the region of interest is also non-constant.

10 . The apparatus of claim 1 , wherein the region of interest has a tapering shape.

11 . The apparatus of claim 10 , wherein the tapering shape has a curvature that is variable.

12 . The apparatus of claim 10 , wherein the tapering shape has a bend that is variable.

13 . The apparatus of claim 1 , wherein the first image comprises an image of an object, and wherein the processing unit is configured to determine whether the image of the object or a bounding box around the image of the object intersects the region of interest.

14 . The apparatus of claim 13 , wherein the processing unit is configured to determine first information indicating a risk of collision if the image of the object or the bounding box around the image of the object intersects the region of interest.

15 . The apparatus of claim 14 , further comprising a second camera configured to view a driver of the vehicle;

wherein the processing unit is configured to determine second information indicating a state of the driver based at least partly on a second image provided by the camera.

16 . The apparatus of claim 15 , wherein the processing unit is configured to determine whether to provide a control signal for operating a device or not based on (1) the first information indicating the risk of collision, and (2) the second information indicating the state of the driver.

17 . A method performed by an apparatus, comprising:

obtaining, by a processing unit, a first image generated by a first camera, wherein the first camera is configured to view an environment outside a vehicle;

determining, by the processing unit, a centerline of a road or a lane in which the vehicle is traveling; and

determining, by the processing unit, a region of interest based on the centerline, the region of interest having a geometry that is based on a feature of the centerline.

18 . The method of claim 17 , wherein the centerline of the road or the lane in which the vehicle is traveling is determined using on a neural network model.

19 . The method of claim 17 , wherein the act of determining the centerline of the road or the lane in which the vehicle is traveling comprises determining a plurality of centerline points.

20 . The method of claim 19 , wherein the centerline points comprise a first centerline point and a second centerline point, and wherein the region of interest is determined based on the first centerline point and the second centerline point.

21 . The method of claim 20 , wherein the act of determining the region of interest comprises:

determining a first left point that is a distance d1 to a left of the first centerline point;

determining a first right point that is the distance d1 to a right of the first centerline point;

determining a second left point that is a distance d2 to a left of the second centerline point; and

determining a second right point that is the distance d2 to a right of the second centerline point.

22 . The method of claim 21 , wherein the act of determining the region of interest further comprises:

determining a left boundary of the region of interest based on the first left point and the second left point; and

determining a right boundary of the region of interest based on the first right point and the second right point.

23 . The method of claim 19 , wherein the centerline points have at least a first centerline point with a first y-coordinate, a second centerline point with a second y-coordinate, and a third centerline point with a third y-coordinate; and

wherein a first difference between the first y-coordinate and the second y-coordinate is different from a second difference between the second y-coordinate and the third y-coordinate.

24 . The method of claim 17 , wherein the act of determining the region of interest comprises determining a left boundary of the region of interest based on the determined centerline, and determining a right boundary of the region of interest based on the determined centerline.

25 . The method of claim 24 , wherein a first distance between the centerline and the left boundary of the region of interest is non-constant, and a second distance between the centerline and the right boundary of the region of interest is also non-constant.

26 . The method of claim 17 , wherein the region of interest has a tapering shape.

27 . The method of claim 26 , wherein the tapering shape has a curvature that is variable.

28 . The method of claim 26 , wherein the tapering shape has a bend that is variable.

29 . The method of claim 17 , wherein the first image comprises an image of an object, and wherein the method further comprises determining whether the image of the object or a bounding box around the image of the object intersects the region of interest.

30 . The method of claim 29 , further comprising determining first information indicating a risk of collision if the image of the object or the bounding box around the image of the object intersects the region of interest.

31 . The method of claim 30 , further comprising a second camera configured to view a driver of the vehicle;

wherein the method further comprises determining second information indicating a state of the driver based at least partly on a second image provided by the camera.

32 . The method of claim 31 , further comprising determining whether to provide a control signal for operating a device or not based on (1) the first information indicating the risk of collision, and (2) the second information indicating the state of the driver.