Table 2.

The architectural configuration of the proposed CAUC model.

Sec	Input	Component/layer	Kernel size	No of comps /filters	Output and its size
Encoder SIDE	Image	IP_Conv	3 × 3	64	E1	224 × 224 × 64
	E1	LCB16	1 × 1, 3 × 3	8, 16	E2	224 × 224 × 16
	E2	Down-Samp1	2 × 2	1	E3	112 × 112 × 16
	E3	Drop Chanel	Dropout_rate = 0.2		E4	112 × 112 × 16
	E4	LCB32	1 × 1, 3 × 3	16, 32	E5	112 × 112 × 32
	E5	Down-Samp2	2 × 2	1	E6	56 × 56 × 32
	E6	Drop Chanel	Dropout_rate = 0.2		E7	56 × 56 × 32
	E7	LCB64	1 × 1, 3 × 3	32, 64	E8	56 × 56 × 64
	E8	Down-Samp3	2 × 2	1	E9	28 × 28 × 64
	E9	Drop Chanel	Dropout_rate = 0.5		E10	28 × 28 × 64
	E10	LCB128	1 × 1, 3 × 3	64, 128	E11	28 × 28 × 128
	E11	Down-Samp3	2 × 2	1	E12	14 × 14 × 128
	E12	Drop Chanel	Dropout_rate = 0.5		E13	14 × 14 × 128
	E13	Middle Layer	3 × 3	256	M	14 × 14 × 256
Decoder side	M	Up-Samp1	2 × 2	1	D1	28 × 28 × 256
	D1	Drop Chanel	Dropout_rate = 0.5		D2	28 × 28 × 256
	E11, M	AG128	1 × 1	128	D3	28 × 28 × 128
	D2, D3	ConCat-1	–	1	D3	28 × 28 × 384
	D3	LCB128	1 × 1, 3 × 3	64, 128	D4	28 × 28 × 128
	D4	Up-Samp2	2 × 2	1	D5	56 × 56 × 128
	D5	Drop Chanel	Dropout_rate = 0.5		D6	56 × 56 × 128
	E8, D4	AG64	1 × 1	64	D7	56 × 56 × 64
	D6, D7	ConCat-2	–	1	D8	56 × 56 × 192
	D8	LCB64	1 × 1, 3 × 3	32, 64	D9	56 × 56 × 64
	D9	Up-Samp3	2 × 2	1	D10	112 × 112 × 64
	D10	Drop Chanel	Dropout_rate = 0.2		D11	112 × 112 × 64
	E5, D9	AG32	1 × 1	32	D12	112 × 112 × 32
	D11,D12	ConCat-3	–	1	D13	112 × 112 × 96
	D13	LCB32	1 × 1, 3 × 3	16, 32	D14	112 × 112 × 32
	D14	Up-Samp4	2 × 2	1	D15	224 × 224 × 32
	D15	Drop Chanel	Dropout_rate = 0.2		D16	224 × 224 × 32
	E2, D14	AG16	1 × 1	32	D17	224 × 224 × 16
	D16,D17	ConCat-4	–	1	D18	224 × 224 × 48
	D18	LCB16	1 × 1, 3 × 3	8. 16	D19	224 × 224 × 16
Feature fusion	D4	CBAM1 + UpSmap	7 × 7, 8 × 8	1	F1	224 × 224 × 16
	D9	CBAM2+ UpSmap	7 × 7, 4 × 4	1	F2	224 × 224 × 64
	D14	CBAM3+ UpSmap	7 × 7, 2 × 2	1	F3	224 × 224 × 32
	D19	CBAM6	7 × 7	1	F4	224 × 224 × 16
	F1, F2, F3, F4	ConCat-5	–	1	F5	224 × 224 × 240
	F5	OP_Conv	3 × 3	5	F6	224 × 224 × 5
	F6	Softmax Layer	–	–	F7	224 × 224 × 5