AI based defect view of road and concrete structure
Innovative solution
The NDT technology was used to conduct by measuring the width of cracks of human visual inspection, and it could measure the area of cracking, whitening, exfoliation by using special software analysis. This innovative technology of AI-based image processing solution provides automatically analysis results of captured images and defects report as quantitative data on the type and amount of damage.
AI-based deep learning technology
In the existing rule-based image analysis technique, in order to detect damages from damage to the exterior of concrete structures, processing of different procedures with different parameters was required according to the type of damage. And the results may vary greatly depending on the shooting quality or the influence of the surrounding shooting environment such as illuminance or focal length. Therefore, this technology improved the morphology technique for structural damage detection and applied the deep learning-based software platform to conduct supervised learning according to the type of damage.
Case study #1
ㆍAI-based deep learning technology
Automatic damage analysis of concrete structures (cracking, whitening, exfoliation, etc.) with AI-based image analysis NDT solution rather than visual inspection
ㆍImage Processing Procedure
Sample Image Acquisition → Pre-processing → Segmentation → Deep Learning → Feature Extraction → Classification → Processing
ㆍAcquired official certificate, satisfies 92.5% of image analysis damage detection result
Case study #2
Road Pavement Auto Damage Analysis Program
Use of package damage analysis program file (automatic classification of collected images)
Analysis engine and original video server are composed of Fiber connected SAN
It save only metadata on the storage and transfer to image file clouding service
Analyze within 10 seconds per every sheet
Provide useful information management by defining OUTPUT calculation criteria such as cumulative damage calculation
Image Analysis using deep learning AI
Image Analysis using deep learning AI
Image Analysis using deep learning AI
Image Analysis using deep learning AI
Image Analysis using deep learning AI
Image Analysis using deep learning AI
Image Analysis using deep learning AI
Image Analysis using deep learning AI
Image Analysis using deep learning AI
Image Analysis using deep learning AI
Image Analysis using deep learning AI
Image Analysis using deep learning AI
Image Analysis using deep learning AI
Image Analysis using deep learning AI
Image Analysis using deep learning AI
Korean
Russian