CNN-Based Material Defect Characterization for Automated Inspection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional methods for identifying and classifying defects in materials during automotive manufacturing are time-consuming and prone to human errors, lacking efficiency and robustness.

Innovation Solution

An AI model, preferably a CNN, is used to identify and characterize defects such as ductile, brittle, and fatigue defects in materials by creating a feature map that classifies these defects based on extracted features from images, utilizing a trained neural network.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If manual identification and classification of defects is used, then human expertise can be applied to understand defect characteristics, but the process becomes highly time consuming and prone to human errors

Engineering Contradiction:
Improvedefect classification accuracyVSAvoidinspection time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent replaces the manual mechanical inspection process with an automated image processing system using convolutional neural networks (CNNs). The system automatically extracts features from material images and classifies defects into categories such as ductile, brittle, and fatigue defects, eliminating the need for manual visual inspection while maintaining high accuracy through trained AI models.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The inspection system performs self-service by automatically analyzing defect characteristics without human intervention. The CNN-based model independently processes images, extracts relevant features, and classifies defects based on learned patterns from training data, enabling the system to serve itself in the inspection task while improving efficiency and consistency.

Inventive Principle:
Principle #25Self-service

2Productivity

If automated image processing is implemented, then inspection speed and consistency are improved, but the system requires complex AI models and training data infrastructure

Engineering Contradiction:
Improveinspection throughputVSAvoidsystem complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent segments the defect classification task into distinct processing stages: image acquisition, feature extraction using CNNs, and classification based on extracted features. This segmentation allows each component to be optimized independently and simplifies the overall system architecture by breaking down the complex AI process into manageable modules that can be implemented and maintained separately.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system performs preliminary action by pre-training CNN models with large datasets of labeled defect images before deployment. This preliminary training phase prepares the AI model to automatically recognize and classify various defect types, reducing the complexity of real-time decision-making during actual inspection operations and enabling faster, more accurate classifications.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4586187A1Method and system for automated characterisation of material defects through an image of the material
Publication Date: 2025.07.16 VALEO ELECTRIFICATION
  • EP4586187A1 patent drawingFigure 1~2
  • EP4586187A1 patent drawingFigure 3~4d
  • EP4586187A1 patent drawingFigure 5a~6c

AI summary

The present disclosure relates to a method for automated characterisation of defects in an image of a material. In particular, it relates to an automated method for identifying and characterizing one or more defects of a material from an image of the material obtained using imaging modalities used, for example, during quality control or defective material analysis in an automotive product manufacturing process.