Automated Crack Detection in Fracture Surfaces
A multi-stage deep learning pipeline for automated fractographic analysis: specimen masking, corner-based orientation normalization, crack and ratchet mark detection, crack contour extraction, and fracture area analysis.
Start New AnalysisSTAGE 1 · 2
Masking & Corners
YOLO isolates the specimen from the background and locates its corners.
STAGE 3
Orientation Normalization
A perspective warp and contrast normalization standardize every specimen into the same canonical view.
STAGE 4
Crack & Ratchet Marks
YOLO locates crack origin sites and ratchet marks on the normalized surface.
STAGE 5
Crack Contour
The crack-growth zone is segmented from the final-fracture zone.
STAGE 6
Fracture Area
Crack-zone and final-fracture areas and their ratio are computed and rolled into a report.