Object Detection in Ground-Penetrating Radar Images Using a Deep Convolutional Neural Network and Image Set Preparation by Migration

Table 1

Accuracy of the cross-validation test using the deep CNN. One line was used for validation images while other lines were used for training images.

The line for validation images	Accuracy	True Positive Ratio (TPR)	True Negative Ratio (TNR)
8009-1	0.945	0.794	0.959
8009-2	0.978	0.719	0.989
8009-3	0.958	0.702	0.991
8009-4	0.979	0.741	0.980