Automated Pavement Crack Damage Detection Using Deep Multiscale Convolutional Features

Table 3

Comparisons of the proposed and other methods on the CrackDataset.

Method	OA	Precision	Recall	-score	mIoU
CrackForest [15]	87.04	86.28	85.46	85.86	59.26
SegNet [37]	96.64	96.86	97.08	96.97	70.56
U-Net [21]	96.58	96.99	97.09	97.04	71.49
PSPNet [38]	96.25	96.90	96.88	96.89	69.63
DeepLabv3+ [39]	96.83	97.01	97.64	97.32	71.77
DeepCrack [30]	97.14	97.33	97.72	97.52	72.04
CrackSeg	98.79	98.00	97.85	97.92	73.53