“Analyzing Images And Videos Of Damaged Vehicles To Determine Damaged Vehicle Parts And Vehicle Asymmetries” in Patent Application Approval Process (USPTO 20240020657): Allstate Insurance Company

2024 FEB 01 (NewsRx) -- By a News Reporter-Staff News Editor at Insurance Daily News -- A patent application by the inventors Campagna, Cory (Deer Park, IL, US); Utke, Jean (Lisle, IL, US), filed on July 17, 2023, was made available online on January 18, 2024, according to news reporting originating from Washington, D.C., by NewsRx correspondents.

This patent application is assigned to Allstate Insurance Company (Northbrook, Illinois, United States).

The following quote was obtained by the news editors from the background information supplied by the inventors: “Aspects of the disclosure relate to analyzing images of vehicles for damage and determining damaged parts of vehicles. Previous approaches involve capturing a small number of images of a damaged vehicle after a damage incident. The images are often taken by a vehicle driver or passenger, who may have difficulty capturing good quality images that accurately represent the damaged portions of the vehicle. When the images do not accurately represent the damaged portions of the vehicle, analysis of the images may be inaccurate. Accordingly, improved methods and systems for collecting data about a vehicle damage incident and analyzing the data are needed.”

In addition to the background information obtained for this patent application, NewsRx journalists also obtained the inventors’ summary information for this patent application: “Aspects of the disclosure provide effective, efficient, scalable, and convenient technical solutions that address and overcome the technical problems associated with analyzing images and videos of damaged vehicles to determine damaged portions of vehicles, damaged parts of vehicles, and whether to repair or replace the damaged portions and/or parts, as well as providing other related functions after a vehicle damage incident.

“In accordance with one or more embodiments, a computing platform having at least one processor, a memory, and a communication interface may receive, via the communication interface, video of a damaged vehicle, perform image analysis of the video to determine one or more frames of the video that include a damaged portion of the vehicle, further analyze the one or more frames of the video that include a damaged portion of the vehicle to determine a damaged cluster of parts of the vehicle, determine whether the damaged cluster of parts should be repaired or replaced, map the damaged cluster of parts to one or more parts in a vehicle-specific database of parts, and generate, based on the mapping, a list of parts for repair or replacement.

“In some embodiments, the computing platform may also determine asymmetry information by comparing an image of the damaged portion of the vehicle to an image of a symmetrical undamaged portion of the vehicle, and determine, based on the asymmetry information, the damaged cluster of parts and an extent of damage to the damaged cluster of parts. In some embodiments, the determination of whether the damaged cluster of parts should be repaired or replaced is based on the extent of damage. In some embodiments, comparing the image of the damaged portion of the vehicle to the image of the symmetrical undamaged portion of the vehicle comprises normalizing a color and/or brightness of the image of the damaged portion of the vehicle and the image of the symmetrical undamaged portion of the vehicle. In some embodiments, comparing the image of the damaged portion of the vehicle to the image of the symmetrical undamaged portion of the vehicle further comprises horizontally flipping and warping the image of the symmetrical undamaged portion of the vehicle.

“In some embodiments, the computing platform may determine an angle of the image of the damaged portion of the vehicle, and determine that the image of the symmetrical undamaged portion of the vehicle is associated with a symmetrical angle.

“In some embodiments, performing image analysis of the video comprises comparing a frame of video to a preceding frame and a following frame to determine one or more motion vectors, and determining whether the frame includes reflections or occlusions based on the one or more motion vectors. In some embodiments, performing image analysis of the video comprises determining a lighting quality of one or more frames of the video.

“In some embodiments, the computing platform determines, based on the list of parts, and based on a local labor rate estimate, a cost to repair the vehicle. The computing platform may also compare the cost to repair the vehicle to a second cost received from a claims adjuster system, and responsive to the cost and the second cost differing by more than a threshold amount, send an alert to the claims adjuster system. The computing platform may further transmit, to a mobile device, a settlement offer based on the cost, and receive, from the mobile device, an acceptance of the settlement offer.

“In some embodiments, the computing platform may select at least one frame, of the one or more frames of the video that depict a damaged portion of the vehicle, as a representative frame of damage, wherein the selection is based on an angle of the at least one frame. The computing platform may further determine a likely-damaged portion of the vehicle based on information received from one or more sensors of the vehicle, and cause a mobile device to display an indication of the likely-damaged portion of the vehicle and an instruction to capture a video of the likely-damaged portion of the vehicle. In some embodiments, the instruction further indicates that the captured video should include a portion of the vehicle that is symmetrical to the likely-damaged portion of the vehicle. The computing platform may further transmit the at least one representative frame to a claims adjuster system, and receive, from the claims adjuster system, a corresponding relevance score for each of the at least one representative frames. The computing platform may further train a model to predict a relevance of an image of a second damaged vehicle using the at least one representative frame and the corresponding relevance score.

“These features, along with many others, are discussed in greater detail below.”

The claims supplied by the inventors are:

“1. A method comprising: receiving, via a communication interface, video of a vehicle; analyzing, by a processor, the video of the vehicle using at least one of an object recognition machine learning model, a damage machine learning model, an angle machine learning model, an image selection machine learning model, a parts cluster machine learning model, or a repair/replace machine model; generating, for the vehicle, using the at least one processor and the one or more machine learning model, an indication that one or more parts for the vehicle are damaged; mapping, by the at least one processor, at least one identifier of a part for the vehicle to the one or more clusters of parts for the vehicle; storing, in a database of the computing platform, mappings of the at least one identifier of a part for the vehicle to the one or more clusters of parts for the vehicle; and transmitting an instruction indicating the at least one identifier.

“2. The method of claim 1, wherein analyzing, by the processor, the video further comprises: generating the one or more of damage information, asymmetry information, or image information based on the video of the vehicle.

“3. The method of claim 2, further comprising: extracting, from the video of the vehicle, images of symmetrical parts; and generating the asymmetry information based on the images of symmetrical parts.

“4. The method of claim 3, further comprising, prior to the extracting: generating reflection data for the images of symmetrical parts; and determining that the reflection data does not indicate reflections in the images.

“5. The method of claim 3, further comprising, prior to the extracting: generating occlusion data for the images of symmetrical parts; and determining that the occlusion data does not indicate occlusions in the images.

“6. The method of claim 3, further comprising, prior to the extracting: generating lighting quality data for the images of symmetrical parts; and determining that the lighting quality data indicates sufficiently good lighting quality for the images.

“7. The method of claim 1, wherein the plurality of vehicles corresponding to the training data comprise vehicles of different makes and models, wherein the clusters of parts are vehicle-independent clusters of parts.

“8. A system comprising: a communication interface; at least one processor; and memory storing instructions that, when executed by the at least one processor, cause the system to: receive, via the communication interface, video of a vehicle; analyze, by a processor, the video of the vehicle using at least one of an object recognition machine learning model, a damage machine learning model, an angle machine learning model, an image selection machine learning model, a parts cluster machine learning model, or a repair/replace machine model; generate, for the vehicle, using the at least one processor and the one or more machine learning model, an indication that one or more parts for the vehicle are damaged; map, by the at least one processor, at least one identifier of a part for the vehicle to the one or more clusters of parts for the vehicle; store, in a database, mappings of the at least one identifier of a part for the vehicle to the one or more clusters of parts for the vehicle; and transmit an instruction indicating the at least one identifier.

“9. The system of claim 8, wherein the instructions, when executed by the at least one processor, further cause the system to: receive, for the vehicle associated with the damage claim, video of the vehicle; and generate the one or more of damage information, asymmetry information, or image information based on the video of the vehicle.

“10. The system of claim 9, wherein the instructions, when executed by the at least one processor, further cause the system to: extract, from the video of the vehicle, images of symmetrical parts; and generate the asymmetry information based on the images of symmetrical parts.

“11. The system of claim 10, wherein the instructions, when executed by the at least one processor, further cause the system to, prior to the extracting: generating reflection data for the images of symmetrical parts; and determining that the reflection data does not indicate reflections in the images.

“12. The system of claim 10, wherein the instructions, when executed by the at least one processor, further cause the system to, prior to the extracting: generating occlusion data for the images of symmetrical parts; and determining that the occlusion data does not indicate occlusions in the images.

“13. The system of claim 10, wherein the instructions, when executed by the at least one processor, further cause the system to, prior to the extracting: generating lighting quality data for the images of symmetrical parts; and determining that the lighting quality data indicates sufficiently good lighting quality for the images.

“14. The system of claim 8, wherein the plurality of vehicles corresponding to the training data comprise vehicles of different makes and models, wherein the clusters of parts are vehicle-independent clusters of parts.

“15. One or more non-transitory computer-readable media storing instructions that, when executed by at least one processor of a system, cause the system to: receive, via the communication interface, video of a vehicle; analyze, by a processor, the video of the vehicle using at least one of an object recognition machine learning model, a damage machine learning model, an angle machine learning model, an image selection machine learning model, a parts cluster machine learning model, or a repair/replace machine model; generate, for the vehicle, using the at least one processor and the one or more machine learning model, an indication that one or more parts for the vehicle are damaged; receive, for a vehicle associated with a damage claim, one or more of damage information, asymmetry information, or image information; generate, for the vehicle, using the at least one processor and the machine learning model, an indication that one or more clusters of parts for the vehicle should be repaired and/or replaced; map, by the at least one processor, at least one identifier of a part for the vehicle to the one or more clusters of parts for the vehicle; store, in a database, mappings of the at least one identifier of a part for the vehicle to the one or more clusters of parts for the vehicle; and transmit an instruction indicating the at least one identifier.

“16. The one or more non-transitory computer-readable media of claim 15, wherein the instructions, when executed by the at least one processor, further cause the system to: receive, for the vehicle associated with the damage claim, video of the vehicle; and generate the one or more of damage information, asymmetry information, or image information based on the video of the vehicle.

“17. The one or more non-transitory computer-readable media of claim 16, wherein the instructions, when executed by the at least one processor, further cause the system to: extract, from the video of the vehicle, images of symmetrical parts; and generate the asymmetry information based on the images of symmetrical parts.

“18. The one or more non-transitory computer-readable media of claim 17, wherein the instructions, when executed by the at least one processor, further cause the system to, prior to the extracting: generating reflection data and/or occlusion data for the images of symmetrical parts; and determining that the reflection data and/or occlusion data does not indicate reflections and/or occlusions in the images.

“19. The one or more non-transitory computer-readable media of claim 17, wherein the instructions, when executed by the at least one processor, further cause the system to, prior to the extracting: generating lighting quality data for the images of symmetrical parts; and determining that the lighting quality data indicates sufficiently good lighting quality for the images.

“20. The one or more non-transitory computer-readable media of claim 15, wherein the plurality of vehicles corresponding to the training data comprise vehicles of different makes and models, wherein the clusters of parts are vehicle-independent clusters of parts.”

URL and more information on this patent application, see: Campagna, Cory; Utke, Jean. Analyzing Images And Videos Of Damaged Vehicles To Determine Damaged Vehicle Parts And Vehicle Asymmetries. U.S. Patent Application Number 20240020657, filed July 17, 2023 and posted January 18, 2024. Patent URL (for desktop use only): https://ppubs.uspto.gov/pubwebapp/external.html?q=(20240020657)&db=US-PGPUB&type=ids

(Our reports deliver fact-based news of research and discoveries from around the world.)