AI Vision Inspection Region Targeting to Reduce Duplicate Checks

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing vision inspection systems often perform duplicate inspections on the same parts across different manufacturing processes, leading to inefficiencies and unnecessary resource utilization.

Innovation Solution

A vision inspection management system utilizing an AI model to identify intensive inspection regions based on process data, enabling optimized vision inspection by intensifying the inspection on specific areas with higher defect probabilities, thereby reducing redundant inspections.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If vision inspection is performed on all areas of a product across multiple manufacturing processes, then defect detection coverage is improved, but inspection efficiency deteriorates due to duplicate inspections

Engineering Contradiction:
Improvedefect detection coverageVSAvoidinspection efficiency
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The patent segments the product inspection into different regions based on defect probability. The AI model divides the product into high-probability regions (requiring intensive inspection) and low-probability regions (requiring standard inspection), allowing differentiated inspection strategies that eliminate redundant inspections while maintaining comprehensive defect detection coverage

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies local quality by assigning different inspection intensities to different regions of the product. High-probability regions receive intensive inspection with higher inspection levels, while low-probability regions receive standard inspection, optimizing resource allocation and eliminating duplicate inspections in areas where defects are unlikely

Inventive Principle:
Principle #3Local quality

2Measurement precision

If intensive inspection is performed on all products, then defect detection accuracy is improved, but resource utilization deteriorates due to unnecessary inspections

Engineering Contradiction:
Improvedefect detection accuracyVSAvoidresource utilization
Core Design Contradiction:
Measurement precisionVSLoss of energy

Solution Approach 1:

The patent applies partial action by performing intensive inspection only on high-probability regions where defects are most likely to occur, rather than applying intensive inspection to the entire product. This partial intensive inspection maintains high defect detection accuracy in critical areas while significantly reducing overall resource consumption

Inventive Principle:
Principle #16Partial or excessive action

Solution Approach 2:

The patent dynamically changes the inspection level parameter based on defect probability. The AI model adjusts the inspection level (e.g., from level 1 to level 3) for different regions based on predicted defect probability, allowing the system to concentrate resources on areas requiring higher accuracy while reducing inspection intensity in low-risk areas

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP3669181B1Vision inspection management method and vision inspection system
Publication Date: 2025.10.01 SAMSUNG ELECTRONICS CO LTD
  • EP3669181B1 patent drawingFigure 1
  • EP3669181B1 patent drawingFigure 2
  • EP3669181B1 patent drawingFigure 3

AI summary

A method, performed by a process management apparatus, of managing vision inspection using an artificial intelligence (AI) model and an apparatus therefor are provided. The method includes obtaining first process data related to a first manufacturing process through which a first object passes, identifying a first region on which intensive inspection is to be performed in an entire region of the first object using the AI model and the first process data, controlling a first vision inspector to inspect the identified first region, and determining whether a defect is present in the identified first region.