Detecting roadway objects in real-time images
The disclosure includes a method that receives a real-time image of a road from a camera sensor communicatively coupled to an onboard computer of a vehicle. The method includes dividing the real-time image into superpixels. The method includes merging the superpixels to form superpixel regions. The method includes generating prior maps from a dataset of road scene images. The method includes drawing a set of bounding boxes where each bounding box surrounds one of the superpixel regions. The method includes comparing the bounding boxes in the set of bounding boxes to a road prior map to identify a road region in the real-time image. The method includes pruning bounding boxes from the set of bounding boxes to reduce the set to remaining bounding boxes. The method may include using a categorization module that identifies the presence of a road scene object in the remaining bounding boxes.
1. A method for identifying roadway objects, the method comprising:
receiving a real-time image of a road from a camera sensor communicatively coupled to an onboard computer of a vehicle;
dividing the real-time image into superpixels;
merging the superpixels to form superpixel regions based on similarity of the superpixels;
drawing initial bounding boxes where each initial bounding box surrounds one of the superpixel regions;
identifying a road region and a central horizon line in the real-time image;
determining a proximity of each of the initial bounding boxes to the road region and the central horizon line in the real-time image;
pruning the initial bounding boxes with a centroid value located above the central horizon line to reduce the initial bounding boxes to remaining bounding boxes;
estimating a size of each of the remaining bounding boxes in the real-time image; and
identifying a set of objects in the road region based on the proximity of the remaining bounding boxes to the road region, the proximity of the remaining bounding boxes to the central horizon line, and the size of each of the remaining bounding boxes.
2. The method of claim 1 , further comprising:
determining a set of categories for one or more objects included in the set of objects.
3. The method of claim 1 , wherein pruning the initial bounding boxes further includes pruning the initial bounding boxes that are outside the road region.
4. The method of claim 1 , further comprising:
generating prior maps from a dataset of road scene images;
wherein identifying the set of objects in the road region based on the proximity of each of the initial bounding boxes to the road region and the size of each of the remaining bounding boxes includes comparing the size of each of the remaining bounding boxes to the prior maps.
5. The method of claim 1 , further comprising:
determining a type of classifier to apply to the initial bounding boxes;
applying the type of classifier to the initial bounding boxes; and
generating a list of object labels for the real-time image based on applying the type of classifier.
6. The method of claim 5 , further comprising:
receiving driving directions; and
modifying the driving directions based on the list of object labels.
7. The method of claim 1 , wherein the real-time image is divided into the superpixels by clustering pixels iteratively using a graph-based agglomerative technique.
8. A non-transitory computer-readable medium having computer instructions stored thereon that are executable by a processing device to perform or control performance of steps comprising:
receiving a real-time image of a road from a camera sensor communicatively coupled to an onboard computer of a vehicle;
dividing the real-time image into superpixels;
merging the superpixels to form superpixel regions based on similarity of the superpixels;
drawing initial bounding boxes where each initial bounding box surrounds one of the superpixel regions;
identifying a road region and a central horizon line;
determining a proximity of each of the initial bounding boxes to the road region and the central horizon line in the real-time image;
pruning the initial bounding boxes with a centroid value located above the central horizon line to reduce the initial bounding boxes to remaining bounding boxes;
estimating a size of each of the remaining bounding boxes in the real-time image; and
identifying a set of objects in the road region based on the proximity of the remaining bounding boxes to the road region, the proximity of the remaining bounding boxes to the central horizon line, and the size of each of the remaining bounding boxes.
9. The non-transitory computer-readable medium of claim 8 , the steps further comprising:
determining a set of categories for one or more objects included in the set of objects.
10. The non-transitory computer-readable medium of claim 9 , the steps further comprising:
determining driving directions based at least in part on the set of categories.
11. The non-transitory computer-readable medium of claim 8 , the steps further comprising:
generating prior maps from a dataset of road scene images;
wherein identifying the set of objects in the road region based on the proximity of each of the initial bounding boxes to the road region and the size of each of the remaining bounding boxes includes comparing the size of each of the remaining bounding boxes to the prior maps.
12. The non-transitory computer-readable medium of claim 8 , the steps further comprising:
determining a type of classifier to apply to the initial bounding boxes;
applying the type of classifier to the initial bounding boxes; and
generating a list of object labels for the real-time image based on applying the type of classifier.
13. The non-transitory computer-readable medium of claim 12 , the steps further comprising:
receiving driving directions; and
modifying the driving directions based on the list of object labels.
14. The non-transitory computer-readable medium of claim 8 , wherein the real-time image is divided into the superpixels by clustering pixels iteratively using a graph-based agglomerative technique.
15. A system comprising:
a memory storing instructions that, when executed by a processor, cause the system to:
receive a real-time image of a road from a camera sensor communicatively coupled to an onboard computer of a vehicle;
divide the real-time image into superpixels;
merge the superpixels to form superpixel regions based on similarity of the superpixels;
draw initial bounding boxes where each initial bounding box surrounds one of the superpixel regions;
identify a road region and a central horizon line;
determine a proximity of each of the initial bounding boxes to the road region and the central horizon line in the real-time image;
prune the initial bounding boxes with a centroid value located above the central horizon line to reduce the initial bounding boxes to remaining bounding boxes;
estimate a size of each of the remaining bounding boxes in the real-time image; and
identify a set of objects in the road region based on the proximity of the remaining bounding boxes to the road region, the proximity of the remaining bounding boxes to the central horizon line, and the size of each of the remaining bounding boxes.
16. The system of claim 15 , wherein the instructions further cause the system to:
determine a set of categories for one or more objects included in the set of objects.
17. The system of claim 16 , wherein the instructions further cause the system to:
determine driving directions based at least in part on the set of categories.
18. The system of claim 15 , wherein the instructions further cause the system to:
generate prior maps from a dataset of road scene images;
wherein identifying the set of objects in the road region based on the proximity of each of the initial bounding boxes to the road region and the size of each of the initial bounding boxes includes comparing the size of each of the initial bounding boxes to the prior maps.
19. The system of claim 15 , wherein the instructions further cause the system to:
determine a type of classifier to apply to the initial bounding boxes;
apply the type of classifier to the initial bounding boxes; and
generate a list of object labels for the real-time image based on applying the type of classifier.
20. The system of claim 19 , wherein the instructions further cause the system to:
receive driving directions; and
modify the driving directions based on the list of object labels.