A risk exposure database contains a compilation of as many building properties or characteristics relevant to insurance as possible. These properties can include characteristics like location coordinates, address, slope, and elevation. Other characteristics include construction type, occupancy type, year built and/or year of renovation, building height, soft stories, number of stories, and floor area. Further characteristics can include roof condition, roof shape, roof covering, roof anchors, roof equipment, cladding, and pounding (distance to closest building). Some of these characteristics can only be assessed by on-site inspections or by official documentation, but others can be measured using visual imagery.
Characteristics addressed in this disclosure include roof shape and roof condition. In one example, roof shapes can be broken into five categories: gambrel roof, gable roof, hipped roof, square roof, and flat roof. Each roof shape has a unique response and damage vulnerability to different natural perils like earthquake or wind.
Deep learning involves computational models composed of multiple processing layers to learn representations of data with multiple levels of abstractions. These models can be thought of as a way to automate predictive analytics. Representation learning is a set of methods that allows a machine to be fed with raw data and to automatically discover the representations needed for detection or classification. Use cases for deep learning include voice recognition, motion detection, translation, and medical diagnosis. By using deep learning algorithms and sample datasets, computers can learn to distinguish and classify a wide range of characteristics to high levels of accuracy, often surpassing the recognition levels of human beings.
One model used for deep learning is the Network in Network model described in the paper “Network In Network” by M. Lin et al. and published in the International Conference on Learning Representations, 2014 (arXiv:1409.1556), the contents of which are hereby incorporated by reference in its entirety. Using the Network in Network model, a number of layers of artificial perception outcomes are generated using micro neural networks with complex structures. The artificial perception outcomes are then stacked and averaged to generate a single global average pooling layer for classification.
When applied to visual recognition, deep learning algorithms can break down an observation (e.g., an image) in a number of different ways to characterize features of the observation. In some examples, deep learning algorithms can be applied to review images as a set of intensity values per pixel, or in a more abstract way as a set of edges, regions of particular shape, etc. Some representations may demonstrate superior performance to others based upon the particular learning task. One of the promises of deep learning is replacing human identification of features with efficient algorithms for unsupervised or semi-supervised feature learning and hierarchical feature extraction.
The inventors recognized that deep learning methodology could be applied to risk exposure database population to analyze aerial imagery and automatically extract characteristics of individual properties, providing fast and efficient automated classification of building styles and repair conditions. In combining location-based vulnerabilities with individual property vulnerabilities identified in part through classification of repair conditions of one or more property features, risk of damage due to disaster can be more accurately estimated.
The forgoing general description of the illustrative implementations and the following detailed description thereof are merely exemplary aspects of the teachings of this disclosure, and are not restrictive.
Systems, methods, and computing system platforms described herein support matching aerial image features of one or more properties to corresponding property conditions (e.g., maintenance levels of property features, including structural features, manmade features included within a vicinity of a structure (e.g., on a property lot, within a property parcel, or a geographic region of the property), and/or natural features included within the vicinity of the structure) through machine learning analysis. In a preferred application, the property condition analysis may be used in estimating damage risk in light of one or more disaster conditions, such as severe storms. The analysis may further aid in estimating costs of repair or replacement of each property, in one example, should a disaster cause the estimated damage. In another example, the analysis may be used to confirm that a property has been repaired.
In one aspect, the present disclosure relates to a method for automatically categorizing a repair condition or maintenance condition of a property characteristic, including obtaining an aerial image of a geographic region including the property; identifying features of the aerial image corresponding to the property characteristic; analyzing the features to determine a property characteristic classification; analyzing a region of the aerial image including the property characteristic to determine a condition classification; and determining, using the property characteristic classification and the condition classification, a risk estimate of damage to the property due to one or more disasters. The property characteristic of a structure such as a house, factory, or barn, in some examples, can include a rooftop, porch, chimney, or skylights. Property characteristics of manmade structures within a property location, in some examples, can include a deck, swimming pool, shed, gazebo, detached garage, tennis court, fence, retaining wall, dock, playground equipment, equipment or vehicles, or greenhouse. Property characteristics of natural features within a property location, in some examples, include trees, ponds, marshes, rivers, lakes, grasses, cliffs, or ocean shore. Further, property characteristic classifications can include shapes, materials, size (breadth and/or height, relative or actual), and/or distance of the property characteristic from other features. In some embodiments, the property characteristic classification may include existing versus not existing (e.g., in the event of determining replacement or removal of a manmade feature at a property location such as a fence). The property characteristic classification, in a particular illustration, may be a rooftop shape. Analyzing the features to determine a property characteristic may include applying a deep learning analysis model to the features. The deep learning analysis model may be NIN.
In some embodiments, analyzing the region of the aerial image including the property characteristic to determine the condition classification includes applying a machine learning analysis model to image pixels within the region. The machine learning analysis model may include a color histogram analysis model. The condition classification may encompass classifications good and bad. Determining the risk estimate may include applying a disaster risk profile corresponding to a first disaster of the at least one disaster and the property characteristic.
In one aspect, the present disclosure relates to a system for automatically categorizing a repair condition of a property characteristic, the system including processing circuitry and a non-transitory computer-readable medium having instructions stored thereon. In some embodiments, the instructions, when executed by the processing circuitry, cause the processing circuitry to obtain an aerial image of a geographic region including the property; identify features of the aerial image corresponding to the property characteristic; analyze the features to determine a property characteristic classification; analyze a region of the aerial image including the property characteristic to determine a condition classification; and determine, using the property characteristic classification and the condition classification, a replacement cost for replacing the property characteristic. The aerial image may be a two-dimensional aerial image.
In some embodiments, the instructions, when executed by the processing circuitry, cause the processing circuitry to, prior to identifying the features, obtain a shape map image including the property; overlay the aerial image with the shape map image; and determine whether a boundary of the property as identified by the shape map matches a boundary of the property as illustrated in the aerial image. The instructions, when executed by the processing circuitry, may cause the processing circuitry to, upon determining that the shape map does not match the boundary of the corresponding property, obtain an alternate aerial image of the property. The instructions, when executed by the processing circuitry, may cause the processing circuitry to, prior to identifying the features, assess orthogonality of the aerial imagery. The property may be a single family home.
In one aspect, the present disclosure relates to a non-transitory computer readable medium having instructions stored thereon, where the instructions, when executed by processing circuitry, cause the processing circuitry to receive identification of a property and at least one property characteristic; obtain an aerial image of a geographic region including the property; and identify respective features of the aerial image corresponding to each property characteristic of the at least one property characteristic. In some embodiments, the instructions, when executed by the processing circuitry, cause the processing circuitry to, for each property characteristic, analyze the corresponding features to determine a respective property characteristic classification, and analyze a region of the aerial image including the respective property characteristic to determine a respective condition classification. The instructions, when executed by the processing circuitry, may cause the processing circuitry to determine, using the property characteristic classification of each property characteristic and the condition classification of each property characteristic, at least one risk estimate representing risk of damage due to disaster.
In some embodiments, the instructions, when executed by the processing circuitry, cause the processing circuitry to, prior to obtaining the aerial image of the geographic region, determine, based upon the at least one property characteristic, a preferred image type corresponding to each property characteristic of the at least one property characteristic. The at least one property characteristic may include two or more property characteristics. The instructions, when executed by the processing circuitry, may cause the processing circuitry to, responsive to determining the preferred image type corresponding to a first property characteristic of the at least one property characteristic is a terrestrial image, obtain a terrestrial image of the geographic region including the property.
In some embodiments, the instructions, when executed by the processing circuitry, cause the processing circuitry to access one or more known property characteristics. Determining the at least one risk estimate may include determining the at least one risk estimate further based on the one or more known property characteristics. The one or more known property characteristics may include at least one of a property age, a property elevation, a property slope, a year built, a year renovated, and a building height.
In some embodiments, receiving identification of the property and the at least one property characteristic includes receiving, via a network from a remote computing device, the identification of the property. The instructions, when executed by the processing circuitry, may cause the processing circuitry to provide, via the network to the remote computing device in real-time responsive to receiving the at least one property characteristic, the at least one risk estimate.
The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate one or more embodiments and, together with the description, explain these embodiments. The accompanying drawings have not necessarily been drawn to scale. Any values dimensions illustrated in the accompanying graphs and figures are for illustration purposes only and may or may not represent actual or preferred values or dimensions. Where applicable, some or all features may not be illustrated to assist in the description of underlying features. In the drawings:
The description set forth below in connection with the appended drawings is intended to be a description of various, illustrative embodiments of the disclosed subject matter. Specific features and functionalities are described in connection with each illustrative embodiment; however, it will be apparent to those skilled in the art that the disclosed embodiments may be practiced without each of those specific features and functionalities.
Reference throughout the specification to “one embodiment” or “an embodiment” means that a particular feature, structure, or characteristic described in connection with an embodiment is included in at least one embodiment of the subject matter disclosed. Thus, the appearance of the phrases “in one embodiment” or “in an embodiment” in various places throughout the specification is not necessarily referring to the same embodiment. Further, the particular features, structures or characteristics may be combined in any suitable manner in one or more embodiments. Further, it is intended that embodiments of the disclosed subject matter cover modifications and variations thereof.
It must be noted that, as used in the specification and the appended claims, the singular forms “a,” “an,” and “the” include plural referents unless the context expressly dictates otherwise. That is, unless expressly specified otherwise, as used herein the words “a,” “an,” “the,” and the like carry the meaning of “one or more.” Additionally, it is to be understood that terms such as “left,” “right,” “top,” “bottom,” “front,” “rear,” “side,” “height,” “length,” “width,” “upper,” “lower,” “interior,” “exterior,” “inner,” “outer,” and the like that may be used herein merely describe points of reference and do not necessarily limit embodiments of the present disclosure to any particular orientation or configuration. Furthermore, terms such as “first,” “second,” “third,” etc., merely identify one of a number of portions, components, steps, operations, functions, and/or points of reference as disclosed herein, and likewise do not necessarily limit embodiments of the present disclosure to any particular configuration or orientation.
Furthermore, the terms “approximately,” “about,” “proximate,” “minor variation,” and similar terms generally refer to ranges that include the identified value within a margin of 20%, 10% or preferably 5% in certain embodiments, and any values therebetween.
All of the functionalities described in connection with one embodiment are intended to be applicable to the additional embodiments described below except where expressly stated or where the feature or function is incompatible with the additional embodiments. For example, where a given feature or function is expressly described in connection with one embodiment but not expressly mentioned in connection with an alternative embodiment, it should be understood that the inventors intend that that feature or function may be deployed, utilized or implemented in connection with the alternative embodiment unless the feature or function is incompatible with the alternative embodiment.
Turning to
The property locations 102b may represent, in some examples, locations of properties insured by a particular insurer, locations of properties recently affected by a disaster such as a tornado, hurricane, earthquake, fire, storm surge, or explosion, locations of properties held in an investment portfolio, or locations of properties considered for addition to an investment portfolio. The property locations 102b may be identified, in some examples, by a street address, global positioning system coordinates, or other geographic coordinates.
In some implementations, the operational flow 100 obtains images (104) at each property location 102b. The images can include aerial imagery 102c stored by the system or accessed by the system from a remote networked location. The aerial imagery 102c, for example, may include a three-dimensional or two-dimensional image of a geographic location including at least one of the property locations 102b. In one example, based upon street address or geographic coordinates, the system may query a remote database to obtain recent aerial imagery 102c including an image of at least one of the property locations 102b. The remote database, in some examples, can include private industry databases such as Google® Earth images by Google, Inc. of Mountain View, Calif. or NTT Geospace Corporation of Japan. In other examples, the remote database can include one or more databases of publicly owned organizations such as the Geospatial Information Authority (GSI) of Japan, the United States Geological Survey, or the Federal Agency for Cartography and Geodesy of Germany. Aerial imagery can be collected from one or more remote network locations, in some embodiments, using an Open Source Geographic Information System (GIS) such as QGIS by the Open Source Geospatial Foundation (OSGeo). The format for the images of the property locations 102b, in some embodiments, depends upon the format accepted by the various sources available for aerial imagery 102c. An example aerial image 510 of multiple property locations 102b is illustrated in
In some embodiments, the system may be configured to query multiple remote database systems to obtain at least two aerial images of a given property location 102b. The aerial images available at the various databases, for example, may differ in resolution and recency of capture. Through collecting two or more images of a particular property, for example, the system may analyze each image to determine a best quality image for use in condition analysis. The condition analysis can include balancing of multiple factors such as, in some examples, clarity, completeness, and recency.
Further, in some embodiments, the system may be configured to query multiple remote database systems to obtain both a two-dimensional aerial image and a three-dimensional aerial image. Different property characteristics may be discerned based upon whether the aerial image is captured in two-dimensional or three-dimensional format. Two-dimensional aerial images, in some examples, can be used to determine location coordinates of the property, street name, occupancy type, floor area, existence of skylights, existence of chimneys, roof condition, roof shape, roof covering, roof anchors, roof equipment, and/or pounding. Three-dimensional aerial images, in comparison, can be used to determine location coordinates, street name, construction type, occupancy type, year built, building height, soft stories, number of stories, roof condition, roof shape, roof covering, roof anchors, roof equipment, cladding, and pounding. Where there is an overlap in characteristics identifiable using either a two-dimensional image or a three-dimensional image, in some embodiments, machine learning analysis of both images can be combined to provide increased confidence in identification of the individual characteristics.
In some implementations, the system obtains (104) shape map images 102a of each of the property locations 102b. The shape map images 102a, for example, include a block shape layout of existing properties in a municipality, such as urban planning maps used in urban planning and development. The shape map images 102a, in another example, can include block shape layout maps used in presenting information to a user of a computer-based navigation system. The shape map images 102a, in one example, can be obtained from Geospatial Information Authority of Japan or Zenrin Co. Ltd. of Japan. An example of an urban planning map 500 is illustrated in
The shape map images 102a, in some embodiments, are used to confirm location of a particular property. The shape map image 102a of a geographical area, for example, can be overlaid with a corresponding aerial image 102c to match properties with images. An example overlaid image 520 of planning map image 500 of
Returning to
As illustrated in
In some implementations, property characteristic profiles 108 are used in classifying property features. The property characteristic profiles 108, for example, may be developed through training the machine learning algorithms using aerial images 102c (and, in some embodiments, terrestrial images 102d) of known property characteristics 110. Each property feature, for example, may be broken down into multiple classifications. In an illustrative example involving a classification of terrestrial images 102d, cladding can include stone, brick, stucco, shingles, vertical boards, horizontal boards, or metal. The machine learning algorithms, for example, may generate a percentage confidence in a match between a new aerial image 102c including a gambrel rooftop and the characteristic profile 108 of gambrel rooftops.
In an illustrative example, turning to
Turning to
To improve accuracy rate, in some implementations, the image can be cropped and/or resized prior to analyzing. For example, the image may be cropped to include the property of interest, or the property of interest plus a portion of its surroundings (e.g., the lot containing the property and/or a portion of the proximate neighborhood in which the property resides). Resizing to a standard image size, in another example, can contribute to accuracy improvement.
Further to roof shape, in some embodiments, feature analysis can be used to discern additional roof features such as, in some examples, roof covering, roof anchors, roof equipment, skylights, widow's walks, turrets, towers, dormers, and/or chimneys. Furthermore, upon identifying the outline of the roof, a footprint of the property location 102b (e.g., size of the roof) can be calculated based upon a scale of the aerial image 102c.
Returning to
In some implementations, terrestrial images 102d may be used to classify features difficult to recognize using aerial imagery 102c. The terrestrial images 102d, in some examples, can include street view images obtained from a street view service or real estate listing. In some examples, street view images including a street-facing view of the property location 102b may be obtained from Google® Street View by Google, Inc., Bing® Maps Streetside by Microsoft Corp. of Redmond, Wash., or Mapillary by Mapillary AB of Sweden. Using a terrestrial image 102d, for example, the system can identify features such as construction type, cladding, building height, number of soft stories, number of floors, location coordinates, street name, slope, elevation, year built, and/or occupancy type. Where the characteristics identifiable via the terrestrial imagery 102d overlap with characteristics identifiable via two-dimensional or three-dimensional aerial imagery 102c, analysis of terrestrial imagery 102d can be combined with analysis of aerial imagery 102c, in some embodiments, to increase confidence in identification of the particular characteristic(s). For example, house siding features may be more easily detected in terrestrial imagery 102d and/or three-dimensional aerial imagery 102c than in two-dimensional aerial imagery 102c.
In some implementations, a condition of each property feature may be classified (112) as a corresponding condition characteristic 116. New properties are in good condition, but property feature conditions can deteriorate over time due to normal wear-and-tear on the property. Further, property features can suffer damage due to external forces such as storms and natural disasters. Eventually, conditions of housing features can deteriorate to the point where repair and/or replacement may be necessary. As with property characteristics described above at block 110, machine learning algorithms can be used to classify a present condition of individual detected property features. Using machine learning for analysis, for example, the system can extract pixel intensity distributions of previously identified property features of the aerial image 102c of the particular property location 102b. In some examples, newly constructed property features generally have sharp contrast and well-defined features in machine learning image analysis. Conversely, weathered or damaged property features can have softened edges, blurred contrasts, and asymmetrical patches of wear. The machine learning classifier used in the machine learning condition analysis, in some embodiments, includes a machine learning analysis to process the aerial image 102c of the particular property location 102b and to classify the condition of previously identified property characteristics 110 as condition characteristics 116. The machine learning analysis, in some examples, can include two-dimensional color histogram analysis or three-dimensional color histogram analysis. In other embodiments, the machine learning analysis may be performed using pattern recognition algorithms (e.g., determining missing fence posts or missing/misaligned rooftop shingles). In other embodiments, the machine learning classifier includes deep learning analysis such as CNN or NIN. For example, CNN analysis may leverage feature maps created during the training process to assess the condition through extracting meaningful features within a given image through convolution and pooling layers. Other machine learning models and algorithms are possible.
In an illustrative example, turning to
Conversely, a set of poor condition aerial images 216a through 220a, are presented alongside corresponding poor condition histograms 216b through 220b. The poor condition histograms 216b through 220b, in an illustrative embodiment, illustrate probability metrics of grayscale color distribution, where a respective maximum value 216c through 220c demonstrates the peak probability of grayscale color distribution of a good condition roof. The maximum values 216c through 220c corresponding to the poor condition rooftops 216a through 220a are markedly lower than corresponding maximum values 210c through 214c of good condition rooftops 210a through 214a. In addition, a width of the distributions of the poor condition histograms 216b through 220b may be markedly wider than a width of the distributions of the good condition histograms 210b through 214b. As can be seen of the aerial images of the poor condition rooftops 216a through 220a, the color distribution is patchy and bleached out, while the edges of the rooftops have lost their crisp lines. Conversely, looking to the aerial images of the good condition rooftops 210a through 214a, the rooftops are more substantially uniform in color with crisp lines on the edges.
Turning to
Although illustrated as a single condition profile 230, individual condition profiles 114 can be generated for each property characteristic (e.g., profiles specific to gabled, gambrel, flat, hipped, square, etc.). In further refinements, in certain embodiments, individual condition profiles 114 can be generated for combinations of property characteristics (e.g., a gabled, shingled roof, a gabled, clay tiled roof, a gabled, metal roof, etc.) to increase accuracy rate based upon pixel densities corresponding to the combined characteristics. Many combinations of characteristics of property features may be used singly or in combination to generate a condition profile 114 designed to accurately identify the condition of the corresponding property feature.
Returning to
In some embodiments, in calculating damage risk, one or more disaster risk profiles 118 can be applied based upon property characteristics 110. Vulnerability to damage due to disaster can vary, in some examples, by location, altitude, slope, rooftop shape, rooftop condition, cladding type, cladding condition, and/or pounding. In a particular example, as shown in
Referring back to
In some implementations, repair or replacement costs can be calculated (126) to determine a cost estimate 128 to place one or more property characteristics in good or “like new” condition. The cost estimates 128, in some embodiments, can be based in part on one or more replacement cost profiles 120. Replacement costs, in some examples, may vary based upon type of materials (e.g., rooftop materials, cladding materials, etc.), size of the job (e.g., estimated area of the roof, estimated length of fence, etc.), and/or additional property characteristics (e.g., contributing to the complexity of the work). The replacement profiles 120, in further examples, may be based in part on location (e.g., varying costs of material and labor regionally). In some implementations, the cost estimates 128 for repair or replacement of the property characteristics may be output to a dashboard interface screen at a remote computing device of a requester, such as an insurance carrier or real estate investment broker.
In some embodiments, cost estimates 128 can be used in automatically analyzing insurance claims. The cost estimate 128, further, may be combined with the risk estimate 122 in analyzing adequacy of insurance coverage for a particular property location 102b. Further, the cost estimates 128 may be used by investors in determining cost of maintaining/upgrading a particular investment property.
Although the operational flow 100 is illustrated as a series of computational stages, in other implementations, more or fewer computational stages may be included. For example, aerial imagery 102c may be analyzed after an insurance claim has been paid to verify that the insured property has been repaired and/or to upgrade the estimated value of the property (e.g., in the circumstance where the owner upgraded the property rather than applying a straight repair/replacement of the damaged property characteristics.
Additionally, in other implementations, certain computational stages may be performed in a different order. For example, cost estimates 128 may be calculated prior to risk estimates 122. Other modifications of the operational flow 100 are possible.
In some implementations, a user of a particular client computing system 306 submits a request to the system 302 through a graphical user interface supplied by a graphical user interface engine 334. The request, for example, may include at least one property identifier 340 as well as one or more property characteristics 342 and/or identification of at least one of a replacement or repair cost estimate, a disaster risk estimate, and a confirmation of repair assessment. The property identifier 340, in some examples, may include location information (e.g., address, geolocation coordinates, lot boundaries, etc.). The location information, for example, may conform with property location information 102b described in relation to
In one example, turning to
In some implementations, the user enters property characteristics associated with each selected property. In other implementations, the characteristics may be accessed upon identification of the property (e.g., selection of a pin 804 identifying a property, as illustrated in
The user interface 820 further presents, in some implementations, premium characteristics 826 such as a replacement value. The premium information, for example, may relate to an insurance policy purchased for the property or parameters for insuring the property (e.g., based upon upgraded or new characteristics). The user, in some embodiments, may be provided to update the premium information via an edit control 828c.
In some implementations, upon identification of property location information, the system 302 accesses imagery of the property from the imagery source computing system(s) 304 using an image acquisition engine 336. The images can include, in some examples, a shape map (e.g., such as the shape maps 102a described in relation to
In some implementations, an image quality analysis and preparation engine 326 analyzes the acquired image(s) to confirm the acquired image(s) contains an adequately clear and detailed image of the property identified by the property identifier 340. In some embodiments, the image quality analysis and preparation engine 326 crops and/or resizes the acquired image(s). For example, the image quality analysis and preparation engine 326 may extract a portion of an acquired image based upon the property (e.g., building size). In some implementations, the image quality analysis and preparation engine 326 resizes each acquired image to a standard size. The standard size, in a particular example, may be 256 by 256 pixels per image. Resizing to a standard size, for example, may increase accuracy in later classification analysis. The image quality analysis and preparation engine 326, in some embodiments, may be configured to apply corrections to the acquired image(s). For example, the image quality analysis and preparation engine 326 may be configured to adjust an aerial image from a normal orthophoto angle, to a true orthophoto version as described above.
In some implementations, the image quality analysis and preparation engine 326 uses shape outlines as part of the analysis. For example, a composite image generation engine 318 may overlay an aerial image with a shape map image, as described in relation to
In some implementations, if the image quality analysis and preparation engine 326 determines that the acquired image is insufficient, the image quality analysis and preparation engine 326 may request a replacement image from the image acquisition engine 336. For example, the image acquisition engine 336 may obtain images based upon a variety of factors including, in some examples, recency of capture, resolution, cost, and/or applicability to a particular property characteristic analysis. Upon determination by the image quality analysis and preparation engine 326 that the first obtained image is insufficient, for example, the image acquisition engine 336 may determine a next best source for obtaining an image of the property.
In some implementations, once an image has been approved (and, optionally, prepared and/or corrected) by the image quality analysis and preparation engine 326, a feature identification engine 320 extracts features from the property image(s) related to the identified property characteristics 342. In the circumstance of a fence, for example, the feature identification engine 320 may identify a perimeter enclosure or partial perimeter enclosure abutting and extending from the property as indicative of a fence.
In some implementations, upon identification of features by the feature identification engine 320, a property characteristic classification engine 322 classifies the property characteristic. For example, the features may be classified using one or more machine learning techniques as described in relation to computational stage 106 (classify features of each image) of
In some implementations, the property characteristic classification engine 322 uses characteristic profile(s) 360 in classifying property characteristics. The property characteristic classification engine 322, for example, may obtain, for each property characteristic being analyzed, a particular characteristic profile 360 from a profile data store 314. The characteristic profiles, for example, may be similar to the characteristic profiles 108 described in relation to
In some implementations, the property characteristic profiles 360 are generated by a characteristic classification learning engine 328. The characteristic classification learning engine 328, for example, may generate characteristic profiles 360 through a sample data set and learning process as described, generally, in relation to
While certain property characteristic classifications are done by analysis, in some implementations, one or more feature classifications may be obtained from previously stored property characteristics 342. For example, based upon tax records, real estate records, etc. some basic materials and structure information regarding a property may be readily available through local and/or remote database system(s). In other examples, the previously stored property characteristics 342 may also be obtained from previous condition assessments performed for a property location by the system 302.
In some implementations, once the property characteristic classification engine 322 has classified the property characteristic(s), a characteristic condition classification engine 324 classifies the condition of each property characteristic. For example, the property characteristics may be classified as described in relation to computational stage 112 (classify condition of each feature) of
In some implementations, the characteristic condition classification engine 324 uses condition profile(s) 362 in classifying property characteristic conditions. The characteristic condition classification engine 324, for example, may obtain, for each property characteristic being analyzed, a particular condition profile 362 from the profile data store 314. The condition profiles 362, for example, may be similar to the condition profiles 114 described in relation to
In some implementations, the condition profiles 362 are generated by a condition classification learning engine 330. The condition classification learning engine 330, for example, may generate condition profiles 362 through a sample data set and learning process as described, generally, in relation to
In some implementations, based upon the output of the property characteristic classification engine 322 and/or the characteristic condition classification engine 324, a risk calculation engine 316 may estimate risk of damage based upon one or more types of disaster. For example, the disaster risk estimate data 352 may be calculated as described in relation to computational stage 120 (calculate damage risk) of
In some embodiments, the risk calculation engine 316 accesses one or more risk profiles 364 from the data store 314 based upon property characteristics 342 and/or condition characteristics 344. The risk profiles 364, for example, may be similar to the disaster risk profiles 118 described in relation to
The risk estimate data 352, in some embodiments, is shared with one or more clients 306. For example, an insurance carrier client 306 may use risk estimate data 352 in making insurance assessments. In another example, a real estate investment broker or firm client 306 may apply risk estimate data 352 when selecting investment properties for an investment portfolio.
Turning to
In some implementations, based upon the output of the property characteristic classification engine 322 and/or the characteristic condition classification engine 324, a cost calculation engine 332 calculates repair or replacement costs for improving problems identified by the characteristic condition classification engine 324 and generates replacement cost estimate data 350. For example, the replacement cost estimate data 350 may be calculated as described in relation to computational stage 126 (calculate repair or replacement costs) of
In some implementations, the cost calculation engine 332 based calculations in part on one or more cost profiles 366 (e.g., similar to the cost profiles 120 of
The cost estimate data 350, in some embodiments, is shared with one or more clients 306. For example, an insurance carrier client 306 may use cost estimate data 350 in determining adequacy of insurance coverage for a particular property. In another example, a real estate investment broker or real estate investment firm client 306 may apply cost estimate data 350 in determining cost of maintaining/upgrading a particular investment property.
In some implementations, the cost estimate data 350 and/or risk estimate data 352 is provided to a requesting client 306, responsive to receiving identification of one or more properties, in real-time or near real-time. For example, a client 306 accessing a dashboard interface, may format a request for a risk estimate or cost estimate related to an identified property within a dashboard interface and submit the request via a network to the system 302. Responsive to the request, the system 302 may conduct the analysis described generally above and respond, in real time or near-real time to the client 306 with a risk analysis or cost analysis. For example, risk analysis information is presented in user interface 840 of
In some implementations, the system may identify, through updated meteorological data, one or more properties within a client real estate portfolio affected by a disaster. The system, further to identification of the affected properties, may proactively prepare a repair estimate to provide to the client corresponding to the real estate portfolio shortly after the disaster took effect. In one example, the Japan Meteorological Agency may update observation data via a web site interface within an hour or so of observation. In this example, the system may provide repair cost estimates to clients between one and two hours after a disaster has taken place.
Rather than sharing cost estimate data 350 and/or risk estimate data 352 directly with the clients 306, in other embodiments, the system 302 may include a report generation engine (not illustrated) that prepares reports regarding condition, damage, and risk assessments of one or more properties. Further, in some embodiments, the system 302 may compare a current condition characteristic 344 to a historic condition characteristic 344 to confirm whether a property owner made repairs to a property (e.g., based upon payment of an insurance claim), replaced one or more features of a property, or removed one or more features of a property (e.g., removed a hazardous collapsing structure, cut back brush encroaching a fire risk distance to a home, etc.). In another illustration, the system 302 may compare a current condition characteristic 344 to a historic condition characteristic 344 to determine whether a repair or replacement is an upgrade (e.g., replacement using superior materials) or a downgrade (e.g., replacing an in-ground pool with an above-ground pool) to the property characteristic. Other modifications of the system 302 are possible.
In some implementations, the method 400 begins with accessing a two-dimensional shape map including the shape of a property (402). The shape map, for example, may be accessed by the image acquisition engine 336 from a shape map imagery source 304, as described in relation to
In some implementations, two-dimensional and/or three-dimensional aerial imagery including an image of the property is accessed (404). The two-dimensional and/or three-dimensional aerial imagery, for example, may be accessed by the image acquisition engine 336 from an aerial imagery source 304, as described in relation to
In some implementations, the aerial imagery is overlaid with the shape map (406). The composite image generation engine 318 of
If a shape match is not identified (408), in some implementations, the method 400 accesses alternate two-dimensional or three-dimensional aerial imagery (410). In some embodiments, the image quality analysis and preparation engine 326 of
If a shape match is identified (408), in some implementations, the formatting of the aerial imagery is prepared (411). In some embodiments, the image is cropped to include the property of interest or the property of interest plus a portion of its surroundings (e.g., the lot containing the property and/or a portion of the proximate neighborhood in which the property resides). Cropping may be based, for example, on the correlation of the shape map to the property. In illustration, the image may be cropped to include the shape map outline of the property plus a border region. In some embodiments, cropping may include cropping to a set shape (e.g., X by X pixels square, X by Y pixels rectangle, etc.). In some embodiments, in addition to cropping the image, the image may be resized. For example, depending upon the resolution of the aerial imagery, the image resolution may be reduced, for example, to fit within an X by X pixels square or an X by Y pixels rectangle. Resizing to a standard image size, for example, can contribute to analysis consistency and improvement in classification accuracy. In further embodiments, the color depth and/or color mapping of the aerial imagery may be adjusted. For consistent color histogram analysis of the imagery, for example, the color depth and color mapping may be made consistent across analyzed images. In a particular example, color images may be converted to grayscale for grayscale image analysis. Adjustments to the imagery, for example, may be effected by the image quality analysis and preparation engine 326, described in relation to
In some implementations, orthogonality of the aerial imagery is assessed (412). For example, the image quality analysis and preparation engine 326 may assess orthogonality of the aerial imagery. The image quality analysis and preparation engine 326, in one example, may determine that orthogonality correction is desired. In another example, if the image represents a normal orthophoto rather than a true orthophoto, the image quality analysis and preparation engine 326 may notify further modules which may compensate for the angle of capture of the aerial image. In other implementations (not illustrated), orthogonality may be assessed without use of an overlaid shape map.
Returning to
In some implementations, one or more property characteristics of the property are classified (418). The property, in some embodiments, is analyzed based on the portion of the aerial image substantially bounded by the shape map of the property. In other embodiments, the property analysis encompasses a surrounding of the property (e.g., features proximate the property upon the property lot, nearby properties, etc.). The property characteristic(s), for example, may be classified by the property characteristic classification engine 322, as described in relation to
In some implementations, if a property characteristic classification is identified (420), a condition of each of the property characteristics can be classified (422). The property characteristic condition(s), for example, may be classified by the characteristic condition classification engine 324, as described in relation to
If, instead, a classification match was not identified (420), in some implementations, alternate two-dimensional or three-dimensional aerial imagery is accessed (410) and the method 400 restarts with overlaying the replacement aerial imagery with the shape map (406). Alternatively, in some embodiments, even if a classification match is not identified, condition may be assessed based upon a default profile associated with the property feature (not illustrated). For example, if the rooftop shape does not match one of the shapes trained into the system, the condition of the rooftop may still be assessed based upon a general condition profile trained with a variety of rooftop shapes. In some examples, the condition of the rooftop may be assessed by comparing the current rooftop conditions to stored historic condition characteristics for the property from a previous condition assessment.
Although the method 400 is illustrated as a series of steps, in other implementations, more or fewer steps may be included. For example, in some implementations, terrestrial images are obtained and used to classify the same and/or different property characteristics, as described in relation to the operational flow 100 of
Additionally, in other implementations, certain steps may be performed in a different order. For example, in some implementations, property characteristics and condition characteristics may be classified (418, 422) in parallel. Other modifications of the operational flow 100 are possible.
Aspects of the present disclosure may be directed to computing systems for categorizing a repair condition of a property characteristic in order to provide dynamic, real-time property condition assessments in response to requests received from users such as insurance carriers or real estate investment brokers using aerial imagery. The implementations described herein improve upon conventional methodologies by applying deep learning analysis models to detected property characteristics of the obtained imagery in order to assess whether a condition of the property characteristics is degraded to a point of needing repair or replacement and determine an amount of risk exposure of the property due to the detected condition of the property characteristics. The implementations described herein improve processing efficiency of the system to reduce an amount of time it takes to perform the condition assessments and automate a condition assessment process.
Next, a hardware description of the computing device, mobile computing device, or server according to exemplary embodiments is described with reference to
Further, a portion of the claimed advancements may be provided as a utility application, background daemon, or component of an operating system, or combination thereof, executing in conjunction with CPU 600 and an operating system such as Microsoft Windows 6, UNIX, Solaris, LINUX, Apple MAC-OS and other systems known to those skilled in the art.
CPU 600 may be a Xenon or Core processor from Intel of America or an Opteron processor from AMD of America, or may be other processor types that would be recognized by one of ordinary skill in the art. Alternatively, the CPU 600 may be implemented on an FPGA, ASIC, PLD or using discrete logic circuits, as one of ordinary skill in the art would recognize. Further, CPU 600 may be implemented as multiple processors cooperatively working in parallel to perform the instructions of the inventive processes described above.
The computing device, mobile computing device, or server in
The computing device, mobile computing device, or server further includes a display controller 608, such as a NVIDIA GeForce GTX or Quadro graphics adaptor from NVIDIA Corporation of America for interfacing with display 610, such as a Hewlett Packard HPL2445w LCD monitor. A general purpose I/O interface 612 interfaces with a keyboard and/or mouse 614 as well as a touch screen panel 616 on or separate from display 610. General purpose I/O interface also connects to a variety of peripherals 618 including printers and scanners, such as an OfficeJet or DeskJet from Hewlett Packard.
A sound controller 620 is also provided in the computing device, mobile computing device, or server, such as Sound Blaster X-Fi Titanium from Creative, to interface with speakers/microphone 622 thereby providing sounds and/or music.
The general purpose storage controller 624 connects the storage medium disk 604 with communication bus 626, which may be an ISA, EISA, VESA, PCI, or similar, for interconnecting all of the components of the computing device, mobile computing device, or server. A description of the general features and functionality of the display 610, keyboard and/or mouse 614, as well as the display controller 608, storage controller 624, network controller 606, sound controller 620, and general purpose I/O interface 612 is omitted herein for brevity as these features are known.
One or more processors can be utilized to implement various functions and/or algorithms described herein, unless explicitly stated otherwise. Additionally, any functions and/or algorithms described herein, unless explicitly stated otherwise, can be performed upon one or more virtual processors, for example on one or more physical computing systems such as a computer farm or a cloud drive.
Reference has been made to flowchart illustrations and block diagrams of methods, systems and computer program products according to implementations of this disclosure. Aspects thereof are implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions/acts specified in the flowchart and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable medium that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable medium produce an article of manufacture including instruction means which implement the function/act specified in the flowchart and/or block diagram block or blocks.
The computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide processes for implementing the functions/acts specified in the flowchart and/or block diagram block or blocks.
Moreover, the present disclosure is not limited to the specific circuit elements described herein, nor is the present disclosure limited to the specific sizing and classification of these elements. For example, the skilled artisan will appreciate that the circuitry described herein may be adapted based on changes on battery sizing and chemistry, or based on the requirements of the intended back-up load to be powered.
The functions and features described herein may also be executed by various distributed components of a system. For example, one or more processors may execute these system functions, where the processors are distributed across multiple components communicating in a network. The distributed components may include one or more client and server machines, which may share processing, as shown on
In some implementations, as illustrated in
The systems described herein may communicate with the cloud computing environment 730 through a secure gateway 732. In some implementations, the secure gateway 732 includes a database querying interface, such as the Google BigQuery platform.
The cloud computing environment 732 may include a provisioning tool 740 for resource management. The provisioning tool 740 may be connected to the computing devices of a data center 734 to facilitate the provision of computing resources of the data center 734. The provisioning tool 740 may receive a request for a computing resource via the secure gateway 732 or a cloud controller 736. The provisioning tool 740 may facilitate a connection to a particular computing device of the data center 734.
A network 702 represents one or more networks, such as the Internet, connecting the cloud environment 730 to a number of client devices such as, in some examples, a cellular telephone 710, a tablet computer 712, a mobile computing device 714, and a desktop computing device 716. The network 702 can also communicate via wireless networks using a variety of mobile network services 720 such as Wi-Fi, Bluetooth, cellular networks including EDGE, 3G and 4G wireless cellular systems, or any other wireless form of communication that is known. In some embodiments, the network 702 is agnostic to local interfaces and networks associated with the client devices to allow for integration of the local interfaces and networks configured to perform the processes described herein.
While certain embodiments have been described, these embodiments have been presented by way of example only, and are not intended to limit the scope of the present disclosures. Indeed, the novel methods, apparatuses and systems described herein can be embodied in a variety of other forms; furthermore, various omissions, substitutions and changes in the form of the methods, apparatuses and systems described herein can be made without departing from the spirit of the present disclosures. The accompanying claims and their equivalents are intended to cover such forms or modifications as would fall within the scope and spirit of the present disclosures.
This application is a continuation of and claims priority to U.S. patent application Ser. No. 17/825,531 entitled “Platform, Systems, and Methods for Identifying Property Characteristics and Property Feature Conditions Through Aerial Imagery Analysis” and filed May 26, 2022, which is a continuation of and claims priority to U.S. patent application Ser. No. 17/518,304 entitled “Platform, Systems, and Methods for Identifying Property Characteristics and Property Feature Conditions Through Aerial Imagery Analysis” and filed Nov. 3, 2021, (now U.S. Pat. No. 11,347,976) which is a continuation of and claims priority to U.S. patent application Ser. No. 16/868,113 entitled “Platform, Systems, and Methods for Identifying Property Characteristics and Property Feature Conditions Through Aerial Imagery Analysis” and filed May 6, 2020, (now U.S. Pat. No. 11,195,058) which is a continuation of and claims priority to U.S. patent application Ser. No. 16/733,888 entitled “Platform, Systems, and Methods for Identifying Property Characteristics and Property Feature Conditions Through Aerial Imagery Analysis” and filed Jan. 3, 2020, (now U.S. Pat. No. 10,650,285) which is a continuation-in-part of and claims priority to U.S. patent application Ser. No. 15/714,376 entitled “Platform, Systems, and Methods for Identifying Property Characteristics and Property Feature Maintenance Through Aerial Imagery Analysis” and filed Sep. 25, 2017, (now U.S. Pat. No. 10,529,029) which claims priority to U.S. Provisional Patent Application Ser. No. 62/398,665, entitled “Platform, Systems, and Methods for Identifying Property Characteristics and Property Feature Maintenance Through Aerial Imagery Analysis,” filed Sep. 23, 2016. All above identified applications are hereby incorporated by reference in their entireties.
Number | Date | Country | |
---|---|---|---|
62398665 | Sep 2016 | US |
Number | Date | Country | |
---|---|---|---|
Parent | 17825531 | May 2022 | US |
Child | 17989991 | US | |
Parent | 17518304 | Nov 2021 | US |
Child | 17825531 | US | |
Parent | 16868113 | May 2020 | US |
Child | 17518304 | US | |
Parent | 16733888 | Jan 2020 | US |
Child | 16868113 | US |
Number | Date | Country | |
---|---|---|---|
Parent | 15714376 | Sep 2017 | US |
Child | 16733888 | US |