Algorithm 1: Simulation for AI-based Crack Width Detection Process

Input: Semantic Segmentation Model Recall and Precision Distribution; Synthetic Inventory

Output: Predicted Crack Width

1: for crack_image in inventory do: 2:   True crack width 🡰 crack_image 3:   Recall 🡰 randomly pick from semantic segmentation model recall distribution 4:   Precision 🡰 randomly pick from semantic segmentation model precision distribution 5:   Predicted crack width 🡰 $\frac{True crack width \ast Recall}{Precision}$