State Farm Mutual Automobile Insurance Company

Ryan Knuffman

Bradley A Sliz

Lucas Allen

Jeremy Carnahan

Bryan R Nussbaum

A computer-implemented method for using a trained generative adversarial network to improve construction and urban planning includes receiving a semantically-segmented point cloud corresponding to a construction site; determining a volumetric soil measurement; and generating a cost estimate. A computing system for using a trained generative adversarial network to improve vehicle orientation and navigation includes one or more processors, and one or more memories having stored thereon computer-executable instructions that, when executed, cause the computing system to: receive a semantically-segmented point cloud corresponding to a construction site; determine a volumetric soil measurement; and generate a cost estimate. A non-transitory computer-readable medium includes computer-executable instructions that, when executed, cause a computer to: receive a semantically-segmented point cloud corresponding to a construction site; determine a volumetric soil measurement; and generate a cost estimate.

Methods and systems for using trained generative adversarial networks to impute 3D data for construction and urban planning

A remotely-controlled (RC) and/or autonomously operated inspection device, such as a ground vehicle or drone, may capture one or more sets of imaging data indicative of at least a portion of an automotive vehicle, such as all or a portion of the undercarriage. The one or more sets of imaging data may be analyzed based upon data indicative of at least one of vehicle damage or a vehicle defect being shown in the one or more sets of imaging data. Based upon the analyzing of the one or more sets of imaging data, damage to the vehicle or a defect of the vehicle may be identified. The identified damage or defect may be compared to a claimed damage or defect to determine whether the claimed damage or defect occurred.

Systems and methods for utilizing machine-assisted vehicle inspection to identify insurance buildup or fraud

A server includes a processor and a memory storing instructions that, when executed by the processor, cause the server to receive two-dimensional (2D) images, analyze the images using a trained deep network to generate points, process the labeled points to identify tie points, and combine the 2D dimensional images into a three-dimensional (3D) point cloud using structure-from-motion. A method for generating a semantically-segmented 3D point cloud from 2D data includes receiving 2D images, analyzing the images using a trained deep network to generate labeled points, processing the points to identify tie points, and combining the 2D images into a 3D point cloud using structure-from-motion. A non-transitory computer readable storage medium stores executable instructions that, when executed by a processor, cause a computer to receive 2D images, analyze the images using a trained deep network to generate labeled points, process the points to identify and combine tie points using structure-from-motion.

Using deep learning and structure-from-motion techniques to generate 3D point clouds from 2D data

A method for using a trained generative adversarial network to improve peril modeling includes receiving a semantically-segmented 3D point cloud; generating a gap-filled point cloud; and generating a digital map. A computing system for using a trained generative adversarial network to improve vehicle orientation and navigation includes one or more processors, and one or more memories having stored thereon computer-executable instructions that, when executed, cause the computing system to: receive a semantically-segmented 3D point cloud; generate a gap-filled point cloud; and generate a digital map. A non-transitory computer-readable medium includes computer-executable instructions that, when executed, cause a computer to: receive a semantically-segmented 3D point cloud; generate a gap-filled point cloud; and generate a digital map.

Methods and systems for using trained generative adversarial networks to impute 3D data for modeling peril

A method includes receiving a navigation data set; generating a combined data set using a trained generative adversarial network; and generating a high resolution map that includes spatial data. A computing system includes: one or more processors, and one or more memories having stored thereon computer-executable instructions that, when executed, cause the computing system to: receive a navigation data set; generate a combined data set using a trained generative adversarial network; and generate a high resolution map that includes spatial data. A non-transitory computer-readable medium includes computer-executable instructions that, when executed, cause a computer to: receive a navigation data set; generate a combined data set using a trained generative adversarial network; and generate a high resolution map that includes spatial data.

Methods and systems for using trained generative adversarial networks to impute 3D data for vehicles and transportation

A method for using a trained generative adversarial network to improve vehicle orientation and navigation includes loading a semantically-segmented 3D point cloud into a virtual reality simulation environment; processing the 3D point cloud; and displaying an output including at least one attribute. A computing system for using a trained generative adversarial network to improve vehicle orientation and navigation includes one or more processors, and one or more memories having stored thereon computer-executable instructions that, when executed, cause the computing system to: load a semantically-segmented 3D point cloud into a virtual reality simulation environment; process the 3D point cloud; and display an output including at least one attribute. A non-transitory computer-readable medium having stored thereon computer-executable instructions that, when executed, cause a computer to: load a semantically-segmented 3D point cloud into a virtual reality simulation environment; process the 3D point cloud; and display an output including at least one attribute.

Methods and systems for using trained generative adversarial networks to impute 3D data for facilities management and operations

A non-transitory computer readable storage medium includes instructions that, when executed by one or more processors, cause a computer to: generate a loss value; update one or more weights of a generative adversarial network; and store the updated weights on a non-transitory computer readable storage medium. A computer-implemented method includes generating a loss value; updating one or more weights of a generative adversarial network; and storing the updated weights on a non-transitory computer readable storage medium. A computing system for training a generative adversarial network includes generating a loss value; updating one or more weights of a generative adversarial network; and storing the updated weights on a non-transitory computer readable storage medium.

Imputation of 3D data using generative adversarial networks

A method for using a trained generative adversarial network to improve underwriting, claim handling and retail operations includes receiving a 3D point cloud; and generating a gap-filled semantically-segmented 3D point cloud using a trained generative adversarial network. A computing system for using a trained generative adversarial network to improve vehicle orientation and navigation includes one or more processors, and one or more memories having stored thereon computer-executable instructions that, when executed, cause the computing system to: receive a 3D point cloud; and generate a gap-filled semantically-segmented 3D point cloud using the trained generative adversarial network. A non-transitory computer-readable medium having stored thereon computer-executable instructions that, when executed, cause a computer to: receive a 3D point cloud; and generate a gap-filled semantically-segmented 3D point cloud using a trained generative adversarial network.

Methods and systems for using trained generative adversarial networks to impute 3D data for underwriting, claim handling and retail operations

A computer-implemented method for labeling a three-dimensional (3D) model using virtual reality (VR) techniques implemented by a computer system including a processor is provided herein. The method may include (i) receiving a 3D model including an environmental feature that is unlabeled, (ii) displaying, through a VR device in communication with the processor, a VR environment to a user representing the 3D model, (iii) prompting a user to input labeling data for the environmental feature displayed within the VR environment of the VR device by prompting the user to select the environmental feature through user interaction with the VR device, and input labeling data for the environmental feature, wherein the labeling data identifies the environmental feature, and (iv) generating a labeled 3D model by embedding the labeling data associated with the selected environmental feature into the 3D model.

Systems and methods for labeling 3D models using virtual reality and augmented reality

A deep artificial neural network (DNN) for generating a semantically-segmented three-dimensional (3D) point cloud is manufactured by a process including obtaining a 3D point cloud, establishing a DNN topology, training the DNN to output labels by subdividing the point cloud, pre-processing the subdivisions, updating weights, and storing weights. Training a DNN includes obtaining a 3D point cloud, establishing a topology of the DNN, training the DNN to output point labels by subdividing, pre-processing the subdivisions, analyzing the features and respective labels of the point cloud to update DNN weights, and storing the weights. A server includes a processor and a memory storing instructions that, when executed by the processor, cause the server to obtain a 3D point cloud, establish a DNN topology, train the DNN to output labels by subdividing, pre-process the subdivisions, analyze the features and respective labels of the point cloud to update weights, and store the weights.

Performing semantic segmentation of 3D data using deep learning

A generative adversarial network (GAN) is manufactured by a process including obtaining a three-dimensional (3D) point cloud, extracting a region from the 3D point cloud, the region corresponding to a gap, analyzing the extracted region to generate a loss, backpropagating the loss, and updating weights of the GAN. A computer-implemented method for training a GAN includes obtaining a 3D point cloud, extracting a region from the 3D point cloud, the region corresponding to a gap, analyzing the extracted region to generate a loss, backpropagating the loss, and updating weights of the GAN. A server includes a processor and a memory storing instructions that, when executed by the processor, cause the server to obtain a 3D point cloud, extract a region from the 3D point cloud, the region corresponding to a gap, analyze the extracted region to generate a loss, backpropagate the loss, and update weights of the GAN.

Growing community of inventors

Danvers, IL, United States of America

Ryan Knuffman