Granted Patent B2

US 12,417,516 · App. 18/377,025 · Granted Sep 16, 2025

Inventors: Yuanming Shu (Toronto, CA); Shuo Tan (Campbell, CA); Hongbo Wang (Hebi, CN)

Assignee: Ecopia Tech Corporation

G06T5/50G06F16/29G06V20/176G06T2207/20221

Abstract

Claims (64)

1. A method comprising:

accessing geospatial imagery that depicts at least one building and its surrounding area;

applying at least one machine learning model to the geospatial imagery to generate artificial parcel data that represents a shape and location of a legal land parcel occupied by the building;

accessing building footprint data comprising at least one polygon that represents a shape and location of the building;

storing the artificial parcel data and the building footprint data in association with address data that represents an address of the building;

receiving a query, the query comprising an address of the building or geospatial coordinates situated within the building footprint or artificial parcel data corresponding to the building; and

providing, in response to the query, the building footprint data and the artificial parcel data corresponding to the building.

2. The method of claim 1 , further comprising:

accessing ground truth parcel data representing shapes and locations of known legal land parcels covered by the geospatial imagery; and

masking the machine learning model from generating artificial parcel data over areas covered by the ground truth parcel data.

3. The method of claim 2 , further comprising:

merging the artificial parcel data with the ground truth parcel data.

4. The method of claim 1 , further comprising:

training the machine learning model to generate the artificial parcel data, wherein training the machine learning model involves training the machine learning model on training data comprising geospatial imagery depicting buildings and the areas surrounding those buildings and ground truth parcel data indicating shapes and locations of legal land parcels occupied by those buildings.

5. The method of claim 4 , wherein training the machine learning model to generate the artificial parcel data comprises:

training the machine learning model to generate a distance-transform raster map over an area covered by the geospatial imagery that comprises a set of pixels each attributed with a distance to a nearest boundary of a legal land parcel.

6. The method of claim 5 , further comprising:

converting the distance-transform raster map into a vector map comprising one or more polygons that represent shapes and locations of boundaries of legal land parcels.

7. The method of claim 6 , wherein converting the distance-transform raster map into the vector map comprises:

performing an instance segmentation process on the distance-transform raster map to generate an instance-labeled raster map; and

converting the instance-labeled raster map into the vector map.

8. The method of claim 1 , wherein:

the geospatial imagery comprises an image tile selected from a grid of image tiles covering a larger geographic area of interest; and

the image tile contains a padding area that overlaps with adjacent image tiles to provide additional context to allow the machine learning model to estimate the shape and location of the legal land parcel.

9. The method of claim 1 , wherein:

receiving the query comprises receiving the query through an application programming interface; and

providing the building footprint data and the artificial parcel data comprises transmitting the building footprint data and the artificial parcel data through the application programming interface.

10. A system comprising one or more computing devices configured to:

access geospatial imagery that depicts at least one building and its surrounding area;

apply at least one machine learning model to the geospatial imagery to generate artificial parcel data that represents a shape and location of a legal land parcel occupied by the building;

access building footprint data comprising at least one polygon that represents a shape and location of the building;

store the artificial parcel data and the building footprint data in association with address data that represents an address of the building;

receive a query, the query comprising an address of the building or geospatial coordinates situated within the building footprint or artificial parcel data corresponding to the building; and

provide, in response to the query, the building footprint data and the artificial parcel data corresponding to the building.

11. The system of claim 10 , wherein the one or more computing devices are further configured to:

train the machine learning model to generate the artificial parcel data, wherein training the machine learning model involves training the machine learning model on training data comprising geospatial imagery depicting buildings and the areas surrounding those buildings and ground truth parcel data indicating shapes and locations of legal land parcels occupied by those buildings.

12. The system of claim 11 , wherein training the machine learning model to generate the artificial parcel data comprises:

training the machine learning model to generate a distance-transform raster map over an area covered by the geospatial imagery that comprises a set of pixels each attributed with a distance to a nearest boundary of a legal land parcel.

13. The system of claim 12 , wherein the one or more computing devices are further configured to:

convert the distance-transform raster map into a vector map comprising one or more polygons that represent shapes and locations of boundaries of legal land parcels.

14. The system of claim 13 , wherein converting the distance-transform raster map into the vector map comprises:

performing an instance segmentation process on the distance-transform raster map to generate an instance-labeled raster map; and

converting the instance-labeled raster map into the vector map.

15. The system of claim 10 , wherein the one or more computing devices are further configured to:

receive the query through an application programming interface; and

transmit the building footprint data and the artificial parcel data in response to the query through the application programming interface.

16. At least one non-transitory machine-readable storage medium comprising instructions that when executed cause one or more processors to:

access geospatial imagery that depicts at least one building and its surrounding area;

apply at least one machine learning model to the geospatial imagery to generate artificial parcel data that represents a shape and location of a legal land parcel occupied by the building;

access building footprint data comprising at least one polygon that represents a shape and location of the building;

store the artificial parcel data and the building footprint data in association with address data that represents an address of the building;

receive a query, the query comprising an address of the building or geospatial coordinates situated within the building footprint or artificial parcel data corresponding to the building; and

provide, in response to the query, the building footprint data and the artificial parcel data corresponding to the building.

17. The at least one non-transitory machine-readable storage medium of claim 16 , wherein the instructions further cause the one or more processors to:

train the machine learning model to generate the artificial parcel data, wherein training the machine learning model involves training the machine learning model on training data comprising geospatial imagery depicting buildings and the areas surrounding those buildings and ground truth parcel data indicating shapes and locations of legal land parcels occupied by those buildings.

18. The at least one non-transitory machine-readable storage medium of claim 16 , wherein training the machine learning model to generate the artificial parcel data comprises:

training the machine learning model to generate a distance-transform raster map over an area covered by the geospatial imagery that comprises a set of pixels each attributed with a distance to a nearest boundary of a legal land parcel.

19. The at least one non-transitory machine-readable storage medium of claim 18 , wherein the instructions further cause the one or more processors to:

convert the distance-transform raster map into a vector map comprising one or more polygons that represent shapes and locations of boundaries of legal land parcels, wherein converting the distance-transform raster map into the vector map comprises:

performing an instance segmentation process on the distance-transform raster map to generate an instance-labeled raster map; and

converting the instance-labeled raster map into the vector map.

20. The at least one non-transitory machine-readable storage medium of claim 16 , wherein the instructions further cause the one or more processors to:

receive the query through an application programming interface; and

transmit the building footprint data and the artificial parcel data in response to the query through the application programming interface.

Continuity (4)

References Cited (35)