Granted Patent B2

US 12,333,801 · App. 18/167,830 · Granted Jun 17, 2025

Inventors: Shantanu Rajora (Fremont, CA); Nitin Das (Haryana, IN)

Assignee: AIDash Inc.

G06V20/188G06T7/0002G06T7/62G06T17/05G06V10/26G06V10/764G06V10/82G06V20/13G06V20/176G06T2207/10032G06T2207/20081G06T2207/20084G06T2207/30184G06T2207/30188

Abstract

Claims (99)

1. A non-transitory computer-readable medium comprising executable instructions, the executable instructions being executable by one or more processors to perform a method, the method comprising:

receiving a set of georeferenced satellite images of a geographic region, the geographic region including multiple geographic areas, the multiple geographic areas including multiple electrical assets of one or more power distribution infrastructures;

generating one or more segmentation maps by providing the set of georeferenced satellite images to one or more fully convolutional networks to classify pixels of the set of georeferenced satellite images as either trees or non-trees;

generating one or more rasters of the geographic region based on the one or more segmentation maps and the set of georeferenced satellite images;

generating one or more vectors based on the one or more rasters, a vector including one or more polygons, a polygon representing a tree and having a set of coordinates defining the polygon;

receiving one or more digital surface models of the geographic region;

generating one or more canopy height models of the geographic region based on the one or more digital surface models by:

downsampling the one or more digital surface models until a peak signal to noise ratio (PSNR) of multiple downsamples of the one or more digital surface models converge to create one or more initial digital terrain models (DTM);

generating one or more initial canopy height models by subtracting the one or more initial digital terrain models from the one or more digital surface models;

determining first pixels in the one or more initial canopy height models that have heights greater than a second threshold to indicate heights of objects above ground height;

interpolating ground height of pixels next to the first pixels to represent ground heights to generate one or more final DTMs; and

generating the one or more canopy height models of the geographic region by subtracting the one or more final digital terrain models from the one or more digital surface models;

associating the heights of the one or more trees with the one or more polygons of the one or more vectors based on the one or more canopy height models;

calculating a first distance between a first polygon having a first height and a first electrical asset of the one or more power distribution infrastructures;

comparing the first distance and the first height;

identifying a first tree represented by the first polygon as a potential hazard based on the comparing of the first distance and the first height; and

providing a notification of the identification of the first tree as a potential hazard.

2. The non-transitory computer-readable medium of claim 1 , the method further comprising:

calculating second distances between second polygons having second heights and second electrical assets of the one or more power distribution infrastructures;

comparing the second distances and the second heights;

identifying second trees represented by the second polygons as potential hazards based on the comparing of the second distances and the second heights;

determining one or more facilities within each of the multiple geographic areas that are connected to second electrical assets of the one or more power distribution infrastructures;

ranking the multiple geographic areas based on the second trees identified as potential hazards and the one or more facilities; and

providing a notification of the ranking of the multiple geographic areas.

3. The non-transitory computer-readable medium of claim 2 , the method further comprising:

modifying at least one vegetation trim plan for the multiple geographic areas based upon the ranking of the multiple geographic areas.

4. The non-transitory computer-readable medium of claim 3 , wherein modifying at least one vegetation trim plan for the multiple geographic areas includes modifying a frequency of occurrence of the at least one vegetation trim plan.

5. The non-transitory computer-readable medium of claim 1 , the method further comprising:

receiving a set of georeferenced stereo satellite images of the geographic region; and

generating the one or more digital surface models of the geographic region based on the set of georeferenced stereo satellite images of the geographic region.

6. The non-transitory computer-readable medium of claim 1 , wherein downsampling the one or more digital surface models until a peak signal to noise ratio (PSNR) of multiple downsamples of the one or more digital surface models converge to create one or more initial digital terrain models (DTM) comprises performing minimum pooling operations on the one or more digital surface models until a calculated peak signal to noise ratio of one or more initial digital terrain models to the one or more digital surface models does not exceed a first threshold,

wherein interpolating the ground height of pixels next to the first pixels to represent ground heights to generate one or more final DTMs comprises: determining second pixels in the one or more initial digital terrain models corresponding to the first pixels and replacing heights of the second pixels with null values, determining third pixels surrounding the second pixels in the one or more initial digital terrain models, and generating one or more final digital terrain models by interpolating heights from the third pixels and replacing the heights of the second pixels with the interpolated heights.

7. The non-transitory computer-readable medium of claim 1 , the method further comprising:

determining a first triangle defined by the first height, a distance from a pixel of the first polygon corresponding to a center of mass of the polygon to the electrical asset, and a calculated hypotenuse; and

utilizing the first triangle in the identification of the first tree represented by the first polygon as a potential hazard.

8. The non-transitory computer-readable medium of claim 1 , wherein the first polygon has one or more heights, and the first height is a maximum height of the one or more heights.

9. The non-transitory computer-readable medium of claim 1 , the method further comprising:

calculating an area of the first polygon; and

calculating a volume of the first tree using the first height and the calculated area of the first polygon.

10. The non-transitory computer-readable medium of claim 9 , the method further comprising:

identifying a species of the first tree; and

utilizing the species of the tree in calculating the volume of the first tree.

11. The non-transitory computer-readable medium of claim 1 , wherein generating the one or more segmentation maps includes:

providing the set of georeferenced satellite images to multiple fully convolutional networks to classify pixels of the set of georeferenced satellite images as either trees or non-trees to obtain multiple intermediate segmentation maps, wherein each of the multiple fully convolutional networks is trained on georeferenced satellite images from a particular satellite image third-party; and

providing the multiple intermediate segmentation maps to a fully convolutional network to combine the multiple intermediate segmentation maps to obtain the one or more segmentation maps.

12. A system comprising at least one processor; and memory containing instructions, the instructions being executable by the at least one processor to:

receive a set of georeferenced satellite images of a geographic region, the geographic region including multiple geographic areas, the multiple geographic areas including multiple electrical assets of one or more power distribution infrastructures;

generate one or more segmentation maps by providing the set of georeferenced satellite images to one or more fully convolutional networks to classify pixels of the set of georeferenced satellite images as either trees or non-trees;

generate one or more rasters of the geographic region based on the one or more segmentation maps and the set of georeferenced satellite images;

generate one or more vectors based on the one or more rasters, a vector including one or more polygons, a polygon representing a tree and having a set of coordinates defining the polygon;

receive one or more digital surface models of the geographic region;

generate one or more canopy height models of the geographic region based on the one or more digital surface models, the one or more canopy height models including heights of one or more trees in the geographic region by the instructions executable by the at least one processor to:

downsample the one or more digital surface models until a peak signal to noise ratio (PSNR) of multiple downsamples of the one or more digital surface models converge to create one or more initial digital terrain models (DTM);

generate one or more initial canopy height models by subtracting the one or more initial digital terrain models from the one or more digital surface models,

determine first pixels in the one or more initial canopy height models that have heights greater than a second threshold to indicate heights of objects above ground height;

interpolate ground height of pixels next to the first pixels to represent ground heights to generate one or more final DTMs; and

generate the one or more canopy height models of the geographic region by subtracting the one or more final digital terrain models from the one or more digital surface models;

associate the heights of the one or more trees with the one or more polygons of the one or more vectors based on the one or more canopy height models;

calculate a first distance between a first polygon having a first height and a first electrical asset of the one or more power distribution infrastructures;

compare the first distance and the first height;

identify a first tree represented by the first polygon as a potential hazard based on the comparing of the first distance and the first height; and

provide a notification of the identification of the first tree as a potential hazard.

13. The system of claim 12 , the instructions being further executable by the at least one processor to:

calculate second distances between second polygons having second heights and second electrical assets of the one or more power distribution infrastructures;

compare the second distances and the second heights;

identify second trees represented by the second polygons as potential hazards based on the comparing of the second distances and the second heights;

determine one or more facilities within each of the multiple geographic areas that are connected to second electrical assets of the one or more power distribution infrastructures;

rank the multiple geographic areas based on the second trees identified as potential hazards and the one or more facilities; and

provide a notification of the ranking of the multiple geographic areas.

14. The system of claim 13 , the instructions being further executable by the at least one processor to modify at least one vegetation trim plan for the multiple geographic areas based upon the ranking of the multiple geographic areas.

15. The system of claim 14 , wherein modifying at least one vegetation trim plan for the multiple geographic areas includes modifying a frequency of occurrence of the at least one vegetation trim plan.

16. The system of claim 12 , the instructions being further executable by the at least one processor to:

receive a set of georeferenced stereo satellite images of the geographic region; and

generate the one or more digital surface models of the geographic region, based on the set of georeferenced stereo satellite images of the geographic region.

17. The system of claim 12 , wherein the instructions executable by the at least one processor to

downsample the one or more digital surface models until a peak signal to noise ratio (PSNR) of multiple downsamples of the one or more digital surface models converge to create one or more initial digital terrain models (DTM) comprises perform minimum pooling operations on the one or more digital surface models until a calculated peak signal to noise ratio of one or more initial digital terrain models to the one or more digital surface models does not exceed a first threshold,

wherein the instructions executable by the at least one processor to interpolate the ground height of pixels next to the first pixels to represent ground heights to generate one or more final DTMs comprises:

determine second pixels in the one or more initial digital terrain models corresponding to the first pixels and replacing heights of the second pixels with null values, determine third pixels surrounding the second pixels in the one or more initial digital terrain models, and generate one or more final digital terrain models by interpolating heights from the third pixels and replacing the heights of the second pixels with the interpolated heights.

18. The system of claim 12 , wherein the first polygon has one or more heights, and the first height is a maximum height of the one or more heights.

19. The system of claim 12 , wherein the instructions executable by the at least one processor to generate the one or more segmentation maps include instructions executable by the at least one processor to:

provide the set of georeferenced satellite images to multiple fully convolutional networks to classify pixels of the set of georeferenced satellite images as either trees or non-trees to obtain multiple intermediate segmentation maps, wherein each of the multiple fully convolutional networks is trained on georeferenced satellite images from a particular satellite image third-party; and

provide the multiple intermediate segmentation maps to a fully convolutional network to combine the multiple intermediate segmentation maps to obtain the one or more segmentation maps.

20. A method comprising:

receiving a set of georeferenced satellite images of a geographic region, the geographic region including multiple geographic areas, the multiple geographic areas including multiple electrical assets of one or more power distribution infrastructures;

generating one or more segmentation maps by providing the set of georeferenced satellite images to one or more fully convolutional networks to classify pixels of the set of georeferenced satellite images as either trees or non-trees;

generating one or more rasters of the geographic region based on the one or more segmentation maps and the set of georeferenced satellite images;

generating one or more vectors based on the one or more rasters, a vector including one or more polygons, a polygon representing a tree and having a set of coordinates defining the polygon;

receiving one or more digital surface models of the geographic region;

generating one or more canopy height models of the geographic region based on the one or more digital surface models, the one or more canopy height models including heights of one or more trees in the geographic region by:

downsampling the one or more digital surface models until a peak signal to noise ratio (PSNR) of multiple downsamples of the one or more digital surface models converge to create one or more initial digital terrain models (DTM);

generating one or more initial canopy height models by subtracting the one or more initial digital terrain models from the one or more digital surface models,

determining first pixels in the one or more initial canopy height models that have heights greater than a second threshold to indicate heights of objects above ground height;

interpolating ground height of pixels next to the first pixels to represent ground heights to generate one or more final DTMs; and

generating the one or more canopy height models of the geographic region by subtracting the one or more final digital terrain models from the one or more digital surface models;

associating the heights of the one or more trees with the one or more polygons of the one or more vectors based on the one or more canopy height models;

calculating a first distance between a first polygon having a first height and a first electrical asset of the one or more power distribution infrastructures;

comparing the first distance and the first height;

identifying a first tree represented by the first polygon as a potential hazard based on the comparing of the first distance and the first height; and

providing a notification of the identification of the first tree as a potential hazard.

Assignments (1)

Continuity (3)

References Cited (30)