1. What defects does AI detection cover?

The AI model is trained on over 200,000 crack defect samples. As the hardest apparent structural defect to detect, cracks already achieve high accuracy; routine apparent defects such as honeycomb, spalling and exposed rebar have relatively obvious features and are easier to detect with higher accuracy.

2. How do detection results flow?

After detection, defects can be mapped onto the 3D model for positioning and filtered by defect type, size and severity level to support standardized report output. AI completes the initial screening, and after professional engineer review the results enter the report, ensuring reliable determination.

3. Applicable scenarios for defect detection

Defect type Detection difficulty Description
Cracks Highest Accuracy above 90%
Honeycomb Lower Routine apparent defect
Spalling Lower Routine apparent defect
Exposed rebar Lower Routine apparent defect

4. Applicable conditions and scenarios

  • Automated screening of apparent bridge defects
  • Detection results enter reports after manual review
  • Covers common defect types of concrete structures

5. Notes

AI detection targets apparent defects; diagnostic conclusions about structural defects must be comprehensively determined by professional engineers per standards.