sam_crack/note.md

Mean

Methods Architecture Parameters (M) GFLOPs Precision (%) Recall (%) F1 (%) IOU (%)
UNet [31] CNN 31.0 54.8 63.9 68.4 66.1 49.3
DeepCrack Y. [13] CNN 14.7 20.1 86.73 57.58 69.2 52.9 DeepCrack Q. [28] CNN 30 137 70.35 70.92 70.6 54.6 TransUNet [34] Transformer 101 48.3 64 67 70.2 56.0 CT CrackSeg [29] Transformer 22.9 41.6 69.1 78 73.3 57.8 VM-UNet* [16],  [30] Mamba 27 4.11 70.7 74.1 72.3 56.7  CrackSegMamba Mamba 0.23 0.70 70.8 75.2 72.9 57.4

Methods	Presision (%)	Recall (%)	F1 (%)	IOU (%)
UNet [31]	63.9	68.4	66.1	49.3
DeepCrack Y. [13]	86.73	57.58	69.2	52.9
DeepCrack Q. [28]	70.35	70.92	70.6	54.6
TransUNet [34]	64	67	70.2	56.0
CT CrackSeg [29]	69.1	78	73.3	57.8
VM-UNet* [16], [30]	70.7	74.1	72.3	56.7
CrackSegMamba	70.8	75.2	72.9	57.4
SAM2 (bbox prompt)	54.14	62.72	53.58	39.60
SAM2 (1 point prompt)	53.85	15.25	12.70	8.43
SAM2 (3 point prompt)	55.26	63.26	45.94	33.35
SAM2 (5 point prompt)	56.38	69.95	51.89	38.50

model, meaniou, stdiou, meanf1, stdf1
bbox, 39.59, 20.43, 53.57, 21.78
1pts, 8.42, 15.3, 12.69, 20.27
3pts, 33.34, 21.83, 45.94, 25.16
5pts, 38.50, 21.47, 51.89, 24.18