Roof Shape Classification Using Segmentation Networks From One Nadir Image

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing systems for determining roof structures from aerial images are computationally expensive, require multiple image types and views, lack camera parameter information, and necessitate manual inspection, making them inefficient and inaccurate.

Innovation Solution

A computer vision system using segmentation networks processes a single nadir aerial image to automatically determine roof structure shapes and ratios, classifying pixels and lines to generate a report without manual intervention.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If multiple image types and views are used for processing, then measurement precision of roof structure is improved, but device complexity increases

Engineering Contradiction:

Improveroof structure identification accuracyVSAvoidsystem complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent extracts and utilizes only the essential nadir aerial image data, discarding the need for multiple image types and views. By focusing on a single nadir image with camera parameter information, the system achieves roof structure identification without the complexity of processing multiple image sources.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent creates a universal system that can identify various roof structure types (gable, hip, flat, etc.) using a single nadir aerial image. The machine learning model is trained to recognize multiple roof configurations from one image type, eliminating the need for multiple specialized image views.

Inventive Principle:

Principle #6Universality (Multi-functionality)

2Measurement precision

If manual inspection is performed to determine roof structure geometries, then measurement precision is improved, but productivity decreases

Engineering Contradiction:

Improveroof structure geometry accuracyVSAvoidprocessing speed

Core Design Contradiction:

Measurement precisionVSProductivity

Solution Approach 1:

The patent replaces manual inspection with an automated machine learning-based image processing system. The neural network model automatically extracts roof structure geometries, identifies roof types, and calculates relevant parameters from nadir aerial images, eliminating the need for human modelers while maintaining accuracy.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system performs self-service by automatically identifying roof structures, determining geometries, and generating reports without human intervention. The machine learning model independently processes images and outputs results, making the system self-sufficient and highly productive.

Inventive Principle:

Principle #25Self-service

3Measurement precision

If a plurality of image types are required for processing, then measurement precision is improved, but loss of time increases

Engineering Contradiction:

Improveroof structure identification accuracyVSAvoidimage processing time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent extracts only the necessary information from a single nadir aerial image, eliminating the time-consuming process of acquiring and processing multiple image types. The system achieves sufficient measurement precision by focusing on essential features visible in one image view.

Inventive Principle:

Principle #2Taking out (Extraction)

4Ease of operation

If camera parameter information is missing, then ease of operation is improved, but measurement precision worsens

Engineering Contradiction:

Improvesystem simplicityVSAvoidaerial image position accuracy

Core Design Contradiction:

Ease of operationVSMeasurement precision

Solution Approach 1:

The patent applies preliminary actions by requiring camera parameter information to be captured and stored during the image acquisition phase. This preliminary data collection enables subsequent precise processing and eliminates the need for complex post-processing to compensate for missing parameters.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS12488485B2Computer vision systems and methods for determining roof shapes from imagery using segmentation networks

Publication Date: 2025.12.02 INSURANCE SERVICES OFFICE INC

AI summary

Computer vision systems and methods for determining roof shapes from imagery using segmentation networks are provided. The system obtains an image of a structure from an image database, and determines a flat roof structure ratio and a sloped roof structure ratio of the roof structure using a neural network. Based on segmentation processing by the neural network, the system determines a flat roof structure ratio and a sloped roof structure ratio based on a portion of the roof structure classified as being flat and a portion of the roof structure classified as being sloped. Then, the system determines a ratio of each shape type of the roof structure using a neural network. The system generates a roof structure shape report indicative of a predominant shape of the roof structure and ratios of each shape type of the roof structure.