Platform, systems, and methods for identifying characteristics and conditions of property features through imagery analysis
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.
1. A system comprising:
a non-transitory computer-readable medium storing an artificial neural network architecture trained to classify building rooftops, the artificial neural network architecture comprising
a first set of layers trained to analyze aerial imagery of a property to extract a plurality of roof features corresponding to a rooftop,
a second set of layers trained to analyze the aerial imagery in view of the plurality of roof features to classify a plurality of roof characteristics, wherein
the plurality of roof characteristics comprises a roof material and a roof type, wherein
a set of roof material classifications for the roof material comprises shingle, metal, and tile, and
a set of roof type classifications for the roof type comprises hip, gable, and flat, and
a third set of layers trained to analyze the aerial imagery in view of the plurality of roof features and plurality of roof characteristics to classify at least one roof condition;
a non-transitory computer-readable data store comprising a plurality of aerial images; and
processing circuitry configured to
access, from the non-transitory computer-readable data store, one or more aerial images of a property parcel,
provide the one or more aerial images for classification by the artificial neural network architecture,
obtain, responsive to providing the one or more aerial images for classification by the artificial neural network architecture, identification of one or more characteristic classifications of the roof and one or more condition classifications of the roof,
receive, via a remote computing device, a request identifying the property parcel; and
responsive to receiving the request, transmit, to the remote computing device, information indicating the one or more characteristic classifications, information indicating the one or more condition classifications, and at least a portion of each image of the one or more aerial images.
2. The system of claim 1 , wherein the processing circuitry is further configured to generate at least one assessment report, wherein the one or more characteristic classifications, the one or more condition classifications, and at least the portion of each image are provided in the at least one assessment report.
3. The system of claim 1 , wherein a set of roof condition classifications comprises a set of condition categories.
4. The system of claim 3 , wherein the set of condition categories comprises a good category and a poor category.
5. The system of claim 3 , wherein the set of roof condition classifications comprises a set of problems.
6. The system of claim 5 , wherein the set of problems comprises at least one of missing shingles or misaligned shingles.
7. The system of claim 1 , wherein:
the artificial neural network architecture comprises a further set of layers trained to analyze the aerial imagery to extract a plurality of structural features indicative of a swimming pool; and
the processing circuitry is configured to obtain, responsive to providing the one or more aerial images to the artificial neural network architecture for classification, indication of presence of the swimming pool.
8. The system of claim 1 , wherein:
the artificial neural network architecture comprises a further set of layers trained to analyze the aerial imagery to extract a second plurality of roof features corresponding to a skylight; and
the processing circuitry is configured to obtain, responsive to providing the one or more aerial images to the artificial neural network architecture for classification, indication of presence of the skylight.
9. The system of claim 1 , wherein:
the artificial neural network architecture comprises a further set of layers trained to analyze the aerial imagery to extract a second plurality of roof features corresponding to a chimney; and
the processing circuitry is configured to obtain, responsive to providing the one or more aerial images to the artificial neural network architecture for classification, indication of presence of the chimney.
10. The system of claim 1 , wherein:
the artificial neural network architecture comprises a further set of layers trained to analyze the aerial imagery to extract a second plurality of natural features corresponding to at least one of a tree, a pond, a marsh, or a river; and
the processing circuitry is configured to obtain, responsive to providing the one or more aerial images to the artificial neural network architecture for classification, indication of presence of one or more of the tree, the pond, the marsh, or the river.
11. The system of claim 1 , wherein the request identifies the property parcel using an address.
12. A method comprising:
accessing, from a non-transitory computer-readable data store comprising a plurality of aerial images, one or more aerial images of a property parcel;
providing, by one or more processors, the one or more aerial images for classification by an artificial neural network architecture trained to classify building rooftops, wherein the artificial neural network architecture comprises
a first set of layers trained to analyze aerial imagery of a property to extract a plurality of roof features corresponding to a rooftop,
a second set of layers trained to analyze the aerial imagery in view of the plurality of roof features to classify a plurality of roof characteristics, wherein
the plurality of roof characteristics comprises a roof material and a roof type, wherein
a set of roof material classifications for the roof material comprises shingle, metal, and tile, and
a set of roof type classifications for the roof type comprises hip, gable, and flat, and
a third set of layers trained to analyze the aerial imagery in view of the roof features and the plurality of roof characteristics to classify at least one roof condition;
responsive to providing the one or more aerial images for classification by the artificial neural network architecture, obtaining, by the one or more processors, one or more characteristic classifications of the roof and one or more condition classifications of the roof;
receiving, via a remote computing device, a request identifying the property parcel; and
responsive to receiving the request, transmitting, to the remote computing device, information indicating the one or more characteristic classifications, information indicating the one or more condition classifications, and at least a portion of each image of the one or more aerial images.
13. The method of claim 12 , further comprising configuring the artificial neural network architecture by:
training the first set of layers of the artificial neural network architecture to analyze aerial imagery to extract the plurality of roof features corresponding to the rooftop;
training the second set of layers of the artificial neural network architecture to analyze the aerial imagery in view of the plurality of roof features to classify the plurality of roof characteristics; and
training the third set of layers trained to analyze the aerial imagery in view of the roof features and the plurality of roof characteristics to classify the at least one roof condition.
14. The method of claim 12 , further comprising generating, by the one or more processors, at least one assessment report, wherein the one or more characteristic classifications, the one or more condition classifications, and at least the portion of each image are provided in the at least one assessment report.
15. The method of claim 12 , wherein the at least one roof condition is classified into one or more roof condition classifications of a set of roof condition classifications, wherein the set of roof condition classifications comprises a set of condition categories.
16. The method of claim 15 , wherein the set of roof condition classifications comprises a set of problems.
17. The method of claim 12 , further comprising:
obtaining, responsive to providing the one or more aerial images to the artificial neural network architecture for classification, indication of presence of a swimming pool;
wherein the artificial neural network architecture comprises a further set of layers trained to analyze the aerial imagery to extract a plurality of structural features indicative of the swimming pool.
18. The method of claim 12 , further comprising:
obtaining, responsive to providing the one or more aerial images to the artificial neural network architecture for classification, indication of presence of a skylight;
wherein the artificial neural network architecture comprises a further set of layers trained to analyze the aerial imagery to extract a second plurality of roof features corresponding to the skylight.
19. The method of claim 12 , further comprising:
obtain, responsive to providing the one or more aerial images to the artificial neural network architecture for classification, indication of presence of a chimney;
wherein the artificial neural network architecture comprises a further set of layers trained to analyze the aerial imagery to extract a second plurality of roof features corresponding to the chimney.
20. The method of claim 12 , further comprising:
obtaining, responsive to providing the one or more aerial images to the artificial neural network architecture for classification, indication of presence of one or more of a tree, a pond, a marsh, or a river;
wherein the artificial neural network architecture comprises a further set of layers trained to analyze the aerial imagery to extract a second plurality of natural features corresponding to at least one of the tree, the pond, the marsh, or the river.