Recognition Method of Tunnel Lining Defects Based on Deep Learning

Table 3

Training results.

Network	Algorithm	Parameter settings	Time (min)	Accuracy (%)
VGG16	SGD		120.33	94.82
VGG16	Adam	,	27.09	94.93
ResNet34	SGD		83.15	93.58
ResNet34	Adam	,	12.35	99.08
Improved ResNet	SGD		87.02	95.42
Improved ResNet	Adam	,	13.07	99.25