Table 4.

The fully convolutional neural networks proposed in Lorenzo et al. (2019) have been trained using a number of datasets with different preprocessing and augmentations.

Setup→	A, A'	B, B'	C, C'	D, D'	E, E'	F, F'	G, G'	H, H'	I, I'	J, J'	K, K'	L, L'	M, M'	N, N'	O, O'
Feature centering	No	Yes	Yes	Yes	Yes	No	Yes	Yes	Yes	Yes	No	Yes	Yes	Yes	Yes
Vertical flip	No	No	Yes	No	Yes	No	No	Yes	No	Yes	No	No	Yes	No	Yes
Horizontal flip	No	No	No	Yes	Yes	No	No	No	Yes	Yes	No	No	No	Yes	Yes
Max. rotation (∡max)	0	0	0	0	0	45	45	45	45	45	90	90	90	90	90
Augmentation factor	1, 2	1, 2	2, 4	2, 4	4, 8	2, 4	2, 4	4, 8	4, 8	8, 16	2, 4	2, 4	4, 8	4, 8	8, 16

In the prime versions, we applied elastic deformations. This table comes from our previous paper (Lorenzo et al., 2019).