Multitype Damage Detection of Container Using CNN Based on Transfer Learning

Table 2

Comparison of accuracy and loss before and after model fine-tuning. The accuracy is the probability of correctly identifying the image and the label; the loss is the difference between the predicted label and the correct label.

Model	Training accuracy		Training loss		Verify accuracy		Verify loss
	Before (%)	Rear (%)	Before (%)	Rear (%)	Before (%)	Rear (%)	Before (%)	Rear (%)
MolileNetV2	86.21	95.32	40.59	23.31	92.92	97.99	28.67	11.21
InceptionV3	82.55	92.85	52.99	27.62	89.41	94.11	35.32	16.11