Platform, systems, and methods for identifying property characteristics and property feature conditions through aerial imagery analysis

• US 10,650,285 B1
• Filed: 01/03/2020
• Issued: 05/12/2020
• Est. Priority Date: 09/23/2016
• Status: Active Grant

• Current Assignee: AON Benfield Incorporated (Aon Plc (Spain))
• Original Assignee: AON Benfield Incorporated (Aon Plc (Spain))
• Inventors: Okazaki, Takeshi
• Primary Examiner(s): Nguyen, Tien C
• Application Number:
  US16/733,888
• Publication Number:
  US 20200143212A1
• Time in Patent Office: 130 Days
• Field of Search: 705 4, 705 5, 705 39, 705 37, 705 38, 705 40, 705 2681, 705 12, 705 272, 235375, 235386, 709225, 707734
• CPC Class Codes: G06F 18/241 relating to the classification model, e.g. parametric or non-parametric approaches
• CPC Class Codes: G06N 20/20 Ensemble learning
• CPC Class Codes: G06N 3/045 Combinations of networks
• CPC Class Codes: G06N 3/08 Learning methods
• CPC Class Codes: G06T 2207/10032 Satellite or aerial image; Remote sensing
• CPC Class Codes: G06T 2207/20076 Probabilistic image processing
• CPC Class Codes: G06T 2207/20081 Training; Learning
• CPC Class Codes: G06T 2207/20084 Artificial neural networks [ANN]
• CPC Class Codes: G06T 2207/30184 Infrastructure
• CPC Class Codes: G06T 7/00 Image analysis
• CPC Class Codes: G06T 7/0002 Inspection of images, e.g. flaw detection
• CPC Class Codes: G06V 10/462 Salient features, e.g. scale invariant feature transforms [SIFT]
• CPC Class Codes: G06V 10/764 using classification, e.g. of video objects
• CPC Class Codes: G06V 10/82 using neural networks
• CPC Class Codes: G06V 20/13 Satellite images
• CPC Class Codes: G06V 20/176 Urban or other man-made structures

Abstract

In an illustrative embodiment, methods and systems for automatically categorizing a condition of a property characteristic may include obtaining aerial imagery of a geographic region including the property, identifying features of the aerial imagery corresponding to the property characteristic, analyzing the features to determine a property characteristic classification, and analyzing a region of the aerial imagery including the property characteristic to determine a condition classification.

Claims

1. A method for automatically categorizing a condition of a characteristic of a property, the method comprising:

obtaining, by processing circuitry, an aerial image of a geographic region including the property;

extracting, by the processing circuitry, one or more features of a plurality of features from the aerial image, wherein

the one or more features are each represented by a set of pixel groupings, and

the one or more features represent the characteristic of the property, wherein

each pixel grouping of the set of pixel groupings comprises at least one of angles, outlines, or substantially homogenous pixel fields;

applying, by the processing circuitry, a first portion of the set of pixel groupings to a first machine learning classifier to determine a characteristic classification of the characteristic from a plurality of potential classifications of the characteristic, wherein

the first machine learning classifier is trained to identify at least a portion of the one or more features from the first portion of the set of pixel groupings; and

applying, by the processing circuitry, a second portion of the set of pixel groupings to a second machine learning classifier to determine a condition classification of the characteristic from a plurality of potential condition classifications of the characteristic, wherein

the second machine learning classifier is trained to identify, from the second portion of the set of pixel groupings, property characteristic conditions of the characteristic having the characteristic classification.

2. The method of claim 1, further comprising comparing, by the processing circuitry, the condition classification to a historic condition classification to identify a change in condition of the characteristic of the property.

3. The method of claim 1, further comprising preparing, by the processing circuitry for presentation to a user, a graphical report comprising condition information regarding the condition classification.

4. The method of claim 3, wherein the preparing comprises preparing the graphical report as an interactive user interface.

5. The method of claim 1, wherein at least one of the first machine learning classifier or the second machine learning classifier comprises a convolutional neural network (CNN) model.

6. The method of claim 1, wherein the plurality of potential condition classifications comprises a numeric scale of condition classifications ranging between a needs replacement condition and a like new condition.

7. The method of claim 1, further comprising determining, by the processing circuitry based in part on the characteristic classification and the condition classification, a cost for repair or replacement of the characteristic of the property.

8. The method of claim 1, further comprising determining, by the processing circuitry based in part on the characteristic classification and the condition classification, a risk estimate of damage to the characteristic of the property due to one or more disasters.

9. A system for automatically categorizing a condition of a property feature of a property, the system comprising:

a non-transitory computer readable storage region storing

a first machine learning analysis model trained to identify one or more property characteristics, and

a second machine learning analysis model trained to identify one or more property conditions;

processing circuitry; and

a non-transitory computer readable medium having instructions stored thereon;

wherein the instructions, when executed by the processing circuitry, cause the processing circuitry to

extract a set of image-related features from an aerial image, wherein the set of image-related features represent the property feature;

apply a first portion of the set of image-related features to the first machine learning analysis model to determine a characteristic classification of the property feature from a plurality of potential characteristic classifications of the property feature; and

apply a second portion of the image-related features to the second machine learning analysis model to determine a condition classification of the property feature from a plurality of potential condition classifications of the property feature, wherein

the second machine learning analysis model is trained to identify, from the second portion of the set of image-related features, property characteristic conditions of the property feature having the characteristic classification.

10. The system of claim 9, wherein each image-related feature of the set of image-related features comprises at least one of angles, outlines, or substantially homogenous pixel fields.

11. The system of claim 9, wherein the instructions, when executed by the processing circuitry, cause the processing circuitry to, prior to extracting the set of image-related features, assess orthogonality of the aerial image.

12. The system of claim 11, wherein the instructions, when executed by the processing circuitry, cause the processing circuitry to, responsive to assessing that the orthogonality of the aerial image does not correspond to a true orthophoto form, cause correction of the aerial image to the true orthophoto form.

13. The system of claim 9, wherein the instructions, when executed by the processing circuitry, cause the processing circuitry to obtain, via a network from a remote computing system, the aerial image of the geographic area.

14. The system of claim 13, wherein the instructions, when executed by the processing circuitry, cause the processing circuitry to, prior to obtaining the aerial image, determine, based at least in part on the property feature, a preferred image type corresponding to the property feature.

15. The system of claim 9, wherein the plurality of potential characteristic classifications comprises at least one of a plurality of potential sizes, a plurality of potential shapes, or a plurality of potential materials of the property feature.

16. A non-transitory computer readable medium having instructions stored thereon, wherein the instructions, when executed by processing circuitry, cause the processing circuitry to:

obtain aerial and/or terrestrial imagery including a property;

extract a plurality of pixel groupings from the aerial and/or terrestrial imagery, wherein

the plurality of pixel groupings includes a respective set of pixel groupings representative of each feature of one or more features of the property, wherein

each pixel grouping of the set of pixel groupings comprises at least one of angles, outlines, or substantially homogenous pixel fields; and

determine at least one characteristic and at least one condition of the property, wherein determining comprises, for each feature of the one or more features of the property,

apply at least a portion of the one or more sets of pixel groupings corresponding to the respective feature to one or more first machine learning classifiers trained to identify one or more characteristics of the respective feature, and

apply at least a portion of the one or more sets of pixel groupings corresponding to the respective feature to one or more second machine learning classifiers trained to identify a condition of the respective feature in view of at least a first characteristic of the one or more characteristics.

17. The non-transitory computer readable medium of claim 16, further comprising using the at least one characteristic and the at least one condition to determine one or more of a risk estimate, a repair estimate, or a replacement estimate.

18. The non-transitory computer readable medium of claim 17, wherein

the instructions, when executed by the processing circuitry, cause the processing circuitry to access one or more known property characteristics, wherein

determining the one or more of the damage risk estimate, the repair estimate, or the replacement estimate comprises applying, to the determining, the one or more known property characteristics.

19. The non-transitory computer readable medium of claim 16, wherein the one or more characteristics of a first feature of the one or more features comprises a shape characteristic and a material characteristic.

20. The non-transitory computer readable medium of claim 16, wherein extracting the plurality of pixel groupings from the aerial and/or terrestrial imagery comprises extracting a first set of pixel groupings from an aerial image and a second set of pixel groupings from a terrestrial image.

21. The non-transitory computer readable medium of claim 16, wherein a first feature of the one or more features comprises a rooftop of a house, and a second feature of the one or more features comprises a structure adjacent to the house.

22. The non-transitory computer readable medium of claim 16, wherein the aerial and/or terrestrial imagery comprises a three-dimensional (3D) aerial image.