Parametric Abnormality Modeling for Synthetic Training Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Limited availability of images of aircraft surfaces with abnormalities hinders the effective training of machine learning algorithms for automated inspection, leading to inconsistent detection and classification of abnormalities, particularly due to restrictions on capturing images and the time-consuming nature of manual data collection.
Innovation Solution
Generating synthetic images of abnormalities using parametric modeling and control points to create a variety of scenarios, including different lighting conditions and viewpoints, allowing for extensive training of machine learning algorithms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If manual inspection is used to detect abnormalities on the surface of a vehicle, then the inspection can be conducted without specialized equipment, but the inspection becomes time consuming and produces inconsistent results
Solution Approach 1:
The patent replaces manual mechanical inspection with an automated image-based inspection system using machine learning algorithms. The system captures images of the vehicle surface and uses trained models to automatically detect and classify abnormalities, eliminating the need for manual surveying while significantly improving inspection speed and consistency.
2Productivity
If automated inspection with machine learning algorithm is used to detect abnormalities, then inspection speed and consistency are improved, but a large number of training images with abnormalities are required
Solution Approach 1:
The patent uses synthetic image generation to create copies of abnormality patterns through parametric modeling. Instead of requiring numerous real images of defective vehicles, the system generates synthetic training images by modeling abnormality surfaces with control points and rendering them onto vehicle surface models, providing ample training data without needing additional physical defect samples.
Solution Approach 2:
The patent employs parametric modeling where abnormality surfaces are defined by adjustable parameters including control point positions, spacing, and surface deformation characteristics. By varying these parameters, the system generates diverse synthetic images of abnormalities with different geometries and positions, creating a comprehensive training dataset from a limited set of base models.
3Reliability
If images of abnormalities are captured manually for training machine learning algorithm, then real defect data is obtained, but the process is time-consuming and limited in quantity
Solution Approach 1:
The patent creates synthetic copies of real abnormality patterns through parametric surface modeling. By capturing the essential geometric characteristics of abnormalities and representing them through control points and surface equations, the system generates realistic training images without requiring physical access to defective vehicles, thus eliminating time-consuming manual data collection while maintaining training effectiveness.
Data Source
AI summary
A computing device, method and computer program product are provided to generate synthetic images of abnormalities on the surface of an object, such as a vehicle. The synthetic images of abnormalities on the surface of an object may be utilized for training a machine learning algorithm to detect and/or classify abnormalities. In the context of a method, a respective abnormality is parametrically modeled by selecting one or more control points that satisfy parameters associated with the respective abnormality and generating a surface representative of the respective abnormality based on the one or more control points. The method also renders a synthetic image of at least a portion of the surface of the object having the respective abnormality as defined by the parametric modeling thereof. The rendering of the synthetic image includes rendering the synthetic image in accordance with a predefined lighting condition and from a predefined viewpoint.