Table 1. Parameter details of the proposed model.

Layer no.	Layer name	Filter shape	No. of filter	Stride	Input1	Input2	Input3	Output
1	Conv-32	3 × 3	32	1	48 × 48 × 1	–	–	48 × 48 × 32
2	Conv-32	3 × 3	32	1	48 × 48 × 32	–	–	48 × 48 × 32
3	Maxpool	2 × 2	--	2	48 × 48 × 32	–	–	24 × 24 × 32
4	Dropout (0.2)	–	–	–	–	–	–	–
5	Conv-64	3 × 3	64	1	24 × 24 × 32	–	–	24 × 24 × 64
6	Conv-64	3 × 3	64	1	24 × 24 × 64	–	–	24 × 24 × 64
7	Conv-64	3 × 3	64	1	24 × 24 × 32	–	–	24 × 24 × 64
8	Conv-64	3 × 3	64	1	24 × 24 × 64	–	–	24 × 24 × 64
9	Concate (4, 6, 8)	–	–	–	24 × 24 × 32	24 × 24 × 64	24 × 24 × 64	24 × 24 × 160
10	Conv-64	1 × 1	64	1	24 × 24 × 160	–	–	24 × 24 × 64
11	Maxpool	2 × 2	–	2	24 × 24 × 64	–	–	12 × 12 × 64
12	Dropout (0.2)	–	–	–	–	–	–	–
13	Conv-128	3 × 3	128	1	12 × 12 × 64	–	–	12 × 12 × 128
14	Conv-128	3 × 3	128	1	12 × 12 × 128	–	–	12 × 12 × 128
15	Conv-128	3 × 3	128	1	12 × 12 × 64	–	–	12 × 12 × 128
16	Conv-128	3 × 3	128	1	12 × 12 × 128	–	–	12 × 12 × 128
17	Concate (12, 14, 16)	–	–	–	12 × 12 × 64	12 × 12 × 128	12 × 12 × 128	12 × 12 × 320
18	Conv-128	1 × 1	128	1	12 × 12 × 320			12 × 12 × 128
19	Maxpool	2 × 2	–	2	12 × 12 × 128	–		6 × 6 × 128
20	Dropout (0.2)	–	–	–	–	–	–	–
21	Conv-256	3 × 3	256	1	6 × 6 × 128	–		6 × 6 × 256
22	Conv-256	3 × 3	256	1	6 × 6 × 256	–		6 × 6 × 256
23	Conv-256	3 × 3	256	1	6 × 6 × 128	–		6 × 6 × 256
24	Conv-256	3 × 3	256	1	6 × 6 × 256	–		6 × 6 × 256
25	Concate (20, 22, 24)	–	–	–	6 × 6 × 128	6 × 6 × 256	6 × 6 × 256	6 × 6 × 640
26	Conv-256	1 × 1	256	1	6 × 6 × 640			6 × 6 × 256
27	Dropout (0.2)	–	–	–	–	–	–	–
28	GAP	–	–	–	6 × 6 × 256	–		256
29	FC	–	–	–	256			512
30	Dropout (0.5)	–	–	–	–	–	–	–
31	FC	–	–	–	512			256
32	Dropout (0.5)	–	–	–	–	–	–	–
33	Softmax	–	–	–	256			7