Skip to main content
. 2022 Oct 26;22(21):8214. doi: 10.3390/s22218214

Table 3.

Comparison of results with other methods.

Method Objtect Accrancy
Ours Semantic Segmentation Defect 96.4%
Cracks 87.83%
Article [37] Bayesian classification Defect 75%
Article [38] YOLOv4 Defect 92.54%
Article [39] k-Means Clustering Defect 90.9%
Article [40] Eddy Current Crack Depth < 2 mm 86%
Crack Depth: 2–6 mm 88%
Article [41] Magnetic Particle Inspection Crack Length: 2.5–5.0 mm 95%
Crack Length: 2–2.5 mm 50%
Crack Length: 1–1.5 mm 30%
Crack Length < 1 mm 0%