Table 3. Summary of all trained networks: loss functions and trainable networks.

Variant name	Comment
C1	Network with GC modules trained with self-supervised multiscale perceptual loss
C2	Network with AGC modules trained with self-supervised single-scale perceptual loss
C3	Network with AGC modules trained with self-supervised multiscale perceptual loss

GC, global context; AGC, adaptive global context.