Table 2.

Average confusion metrics obtained by six CNN models included in DUNEScan.

	True negative (%)	False negative (%)	False positive (%)	True positive (%)
ResNet50	78.83	21.16	13.08	86.91
EfficientNet	79.27	20.72	13.67	86.32
Inceptionv3	74.29	25.7	7.47	92.52
MobileNetv2	75.31	24.68	13.51	86.48
BYOL	69.53	30.46	11.18	88.81
SwAV	70.7	29.29	11.52	88.47