Table 1.

Computational complexity (Training and Testing) of different CNN-based deep learning architectures.

Networks	# Params	Training time(s) / epoch	Test time(ms)	Total epoch
Conv AE	27,891,584	21 ± 1	19	10
UN - 1 layer	403,328	8 ± 1	19	10
UN - 2 layers	1861,504	14 ± 1	20	10
UN	31,025,024	27 ± 2	20	10
UN++	36,158,083	30 ± 1	21	10
Dense-UN	38,961,027	48 ± 2	21	10
Res-UN	33,158,351	33 ± 2	21	30
Att-UN	37,334,803	37 ± 1	21	30
Att Res-UN	39,090,515	48 ± 3	22	30
R2-UN	176,175,362	130 ± 3	24	10
Double UN	3760,899	25 ± 1	19	10