Concrete Crack Image Review Using Difficulty-Based Region Highlighting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for detecting distresses in concrete structures, such as cracks, suffer from false or missed detections due to image quality issues, making it burdensome for users to correct or edit the data accurately.
Innovation Solution
An information processing apparatus that detects candidate regions in images containing multiple crack endpoints and calculates a difficulty score based on the number and attributes of cracks, highlighting regions requiring user correction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the user checks and corrects each detected distress individually, then the accuracy of distress data can be improved, but the time and effort required for correction increases significantly
Solution Approach 1:
The patent segments the image into multiple regions and groups distresses by spatial proximity, creating clusters that can be processed together rather than individually. This segmentation allows the system to present grouped distress information to the user, reducing the number of individual correction actions required while maintaining detection accuracy.
Solution Approach 2:
The system performs preliminary grouping and organization of distresses by region and proximity before presenting them to the user for correction. By pre-processing and organizing the distress data into logical clusters, the system reduces the cognitive load and time required for user correction, as users can review and correct multiple related distresses in a single action rather than individually.
2Reliability
If all detected distresses are presented to the user for verification, then the reliability of distress data can be improved, but the complexity of the interface and user task increases
Solution Approach 1:
The interface is segmented by dividing distresses into region-based groups, allowing users to verify distresses in manageable portions rather than viewing all distresses simultaneously. This segmentation reduces interface complexity while maintaining data reliability through systematic verification.
Solution Approach 2:
The system applies different presentation qualities to different distress groups based on their characteristics, such as highlighting high-confidence detections versus low-confidence ones, or grouping by severity. This local differentiation allows the interface to present information in a more intuitive and less complex manner while ensuring thorough verification.
3Extent of automation
If the system divides images into predetermined rectangles for crack detection, then the automation of distress detection can be improved, but the precision of distress location and classification may deteriorate
Solution Approach 1:
The system transitions from static predetermined rectangular divisions to dynamic region segmentation based on actual distress locations and spatial relationships. By dynamically adjusting region boundaries to match the actual distribution of distresses, the system maintains high automation while improving location precision, as regions are adapted to the specific characteristics of each image rather than using fixed grids.
Data Source
AI summary
An information processing apparatus includes one or more memories storing instructions and one or more processors. The one or more processors are configured to, upon executing the instructions, detect, as a candidate region, a region within a predetermined size range including two or more end points of distresses from a captured image of an inspection target, acquire a difficulty in a case where a user determines whether to perform a predetermined task on the distresses having the end points included in the candidate region, and display the candidate region based on the difficulty.