Technologies for using image data analysis to assess and classify hail damage
Systems and methods for analyzing image data to assess property damage are disclosed. According to certain aspects, a server may analyze segmented digital image data of a roof of a property using a convolutional neural network (CNN). The server may extract a set of features from a set of regions output by the CNN. Additionally, the server may analyze the set of features using an additional image model to generate a set of outputs indicative of a confidence level that actual hail damage is depicted in the set of regions.
1 . A computer-implemented method, comprising:
accessing a digital image;
identifying, by a processor, and using a convolutional neural network (CNN), a region of potential damage depicted in the digital image;
identifying, by the processor, a first shape feature associated with a first section of the digital image, and a second shape feature associated with a second section of the digital image, wherein:
the first section is disposed within the region of potential damage, and
the second section is disposed outside of the region of potential damage;
determining, by the processor, that a first curvature of the first shape feature is greater than a second curvature of the second shape feature; and
generating, by the processor, using a classification model, and based on determining that the first curvature of the first shape feature is greater than the second curvature of the second shape feature, an output indicating a presence of damage associated with the region of potential damage, wherein the output is used to determine an estimated damage amount.
2 . The computer-implemented method of claim 1 , wherein identifying the first shape feature comprises extracting the first shape feature from the digital image based on an aspect ratio of the first shape feature.
3 . The computer-implemented method of claim 1 , further comprising:
training the CNN using training data comprising training images and training labels, wherein the trained CNN is configured to:
classify the digital image based on at least one of the training images and the training labels, and
generate an image depicting the region of potential damage.
4 . The computer-implemented method of claim 3 , wherein:
the training images comprise a first image that depicts a type of damage and a second image that does not depict the type of damage; and
the training labels comprise:
first data identifying a portion of the first image as a first region depicting the type of damage, and
second data identifying a remaining portion of the first image as a second region not depicting the type of damage, the remaining portion excluding the first region depicting the type of damage.
5 . The computer-implemented method of claim 1 , wherein accessing the digital image comprises segmenting the digital image using a sliding window technique.
6 . The computer-implemented method of claim 1 , wherein the output comprises a binary output indicating that damage is represented in the first section of the digital image.
7 . The computer-implemented method of claim 1 , wherein generating the output using the classification model comprises:
inputting, by the processor, the first shape feature into the classification model; and
assigning a confidence level to the output based on a likelihood of the first shape feature indicating the presence of damage in the digital image.
8 . The computer-implemented method of claim 7 , wherein generating the output using the classification model further comprises determining the output based on whether the confidence level meets or exceeds a damage threshold value.
9 . A system for analyzing image data to automatically assess damage to a property, comprising:
a memory configured to store non-transitory computer executable instructions; and
a processor interfacing with the memory, and configured to execute the non-transitory computer executable instructions to cause the processor to perform operations comprising:
accessing a digital image;
identifying, using a convolutional neural network (CNN), a region of potential damage depicted in the digital image;
identifying a first shape feature associated with a first section of the digital image and a second shape feature associated with a second section of the digital image, wherein:
the first section is disposed within the region of potential damage, and
the second section is disposed outside of the region of potential damage;
determining that a first contour curvature of the first shape feature is greater than a second contour curvature of the second shape feature; and
generating, using a classification model, and based at least in part on determining that the first contour curvature of the first shape feature is greater than the second contour curvature of the second shape feature, an output indicating a presence of damage associated with the region of potential damage, wherein the output is used to determine an estimated damage amount.
10 . The system of claim 9 , wherein accessing the digital image comprises receiving the digital image from an unmanned aerial vehicle.
11 . The system of claim 9 , wherein generating the output using the classification model comprises inputting the first shape feature into the classification model to generate a binary output indicating whether damage is represented in the first shape feature.
12 . The system of claim 9 , wherein the output is further based on at least one of an aspect ratio of the first section of the digital image or a pixel area of the first section of the digital image.
13 . The system of claim 9 , wherein generating the output using the classification model comprises:
inputting the first shape feature into the classification model; and
generating a confidence level based on a likelihood of the first shape feature indicating the presence of damage in the digital image.
14 . The system of claim 13 , wherein generating the output using the classification model further comprises determining the output based on whether the confidence level meets or exceeds a damage threshold value.
15 . A non-transitory computer-readable storage medium configured to store instructions, the instructions when executed by a processor causing the processor to perform operations comprising:
accessing a digital image;
identifying, using a convolutional neural network (CNN), a region of potential damage depicted in the digital image;
identifying a first shape feature associated with a first section of the digital image and a second shape feature associated with a second section of the digital image, wherein:
the first section is disposed within the region of potential damage, and
the second section is disposed outside of the region of potential damage;
determining that a first contour curvature of the first shape feature is greater than a second contour curvature of the second shape feature; and
generating, using a classification model, and based at least in part on determining that the first contour curvature of the first shape feature is greater than the second contour curvature of the second shape feature, an output indicating a presence of damage associated with the region of potential damage, wherein the output is used to determine an estimated damage amount.
16 . The non-transitory computer-readable storage medium of claim 15 , wherein the digital image accessed from a set of digital images determined based on cropping digital image data into the set of digital images.
17 . The non-transitory computer-readable storage medium of claim 15 , wherein the output comprises a confidence level associated with the presence of damage in the digital image.
18 . The non-transitory computer-readable storage medium of claim 15 , wherein the output is further based on at least one of an aspect ratio of the first section of the digital image or a pixel area of the first section of the digital image.
19 . The non-transitory computer-readable storage medium of claim 15 , wherein accessing the digital image comprises segmenting digital image data into a set of digital images comprising the digital image.
20 . The non-transitory computer-readable storage medium of claim 19 , wherein accessing the digital image further comprises receiving the digital image data from an unmanned aerial vehicle.