Table 4.

TrIncNet model’s implementation details.

Layer No.			Layer Name	Input Shape	Connected to	Output Shape	Parameters
1			Input Layer	$256\times 256\times 3$	–	$256\times 256\times 3$	$0$
2			Patches	$256\times 256\times 3$	Input Layer	$256\times 768$	$0$
3			Patch Encoder	$256\times 768$	Patches	$256\times 256$	$262400$
4			Layer Normalization #1	$256\times 256$	Patch Encoder	$256\times 256$	$512$
5			Multi-Head Attention #1	$256\times 256$	Layer Normalization #1	$256\times 256$	$3155200$
6			Add #1	$256\times 256$ , $256\times 256$	Multi-Head attention #1, Patch Encoder	$256\times 256$	0
7			Layer Normalization #2	$256\times 256$	Add #1	$256\times 256$	$512$
8	Inception Module #1		Reshape #1	$\mathbf{256}\times \mathbf{256}$	Layer Normalization #2	$\mathbf{16}\times \mathbf{16}\times \mathbf{256}$	$\mathbf{0}$
			Conv2D #1	$\mathbf{16}\times \mathbf{16}\times \mathbf{256}$	Reshape #1	$\mathbf{16}\times \mathbf{16}\times \mathbf{96}$	$\mathbf{24672}$
			Conv2D #2	$\mathbf{16}\times \mathbf{16}\times \mathbf{256}$	Reshape #1	$\mathbf{16}\times \mathbf{16}\times \mathbf{16}$	$\mathbf{4112}$
			Conv2D #3	$\mathbf{16}\times \mathbf{16}\times \mathbf{256}$	Reshape #1	$\mathbf{16}\times \mathbf{16}\times \mathbf{64}$	16448
			MaxPooling2D #1	$\mathbf{16}\times \mathbf{16}\times \mathbf{256}$	Reshape #1	$\mathbf{16}\times \mathbf{16}\times \mathbf{256}$	0
			Conv2D #4	$\mathbf{16}\times \mathbf{16}\times \mathbf{96}$	Conv2D #1	$\mathbf{16}\times \mathbf{16}\times \mathbf{128}$	$\mathbf{110720}$
			Conv2D #5	$\mathbf{16}\times \mathbf{16}\times \mathbf{16}$	Conv2D #2	$\mathbf{16}\times \mathbf{16}\times \mathbf{32}$	$\mathbf{12832}$
			Conv2D #6	$\mathbf{16}\times \mathbf{16}\times \mathbf{256}$	MaxPooling2D #1	$\mathbf{16}\times \mathbf{16}\times \mathbf{32}$	$\mathbf{8224}$
			Concatenate #1	$\mathbf{16}\times \mathbf{16}\times \mathbf{64},$ $\mathbf{16}\times \mathbf{16}\times \mathbf{128}$ , $\mathbf{16}\times \mathbf{16}\times \mathbf{32}$ , $\mathbf{16}\times \mathbf{16}\times \mathbf{32}$	Conv2D #3, Conv2D #4, Conv2D #5, Conv2D #6	$\mathbf{16}\times \mathbf{16}\times \mathbf{256}$	0
			Reshape #2	$\mathbf{16}\times \mathbf{16}\times \mathbf{256}$	Concatenate #1	$\mathbf{256}\times \mathbf{256}$	0
9			Add #2	$256\times 256$ , $256\times 256$	Reshape #2, Add #1	$256\times 256$	0
10			Add #3	$256\times 256$ , $256\times 256$	Add #2, Patch Encoder	$256\times 256$	0
11			Layer Normalization #3	$256\times 256$	Add #3	$256\times 256$	$512$
12			Multi-Head attention #2	$256\times 256$	Layer Normalization #3	$256\times 256$	$3155200$
13			Add #4	$256\times 256$ , $256\times 256$	Multi-Head attention #2, Add #3	$256\times 256$	0
14			Layer Normalization #4	$256\times 256$	Add #4	$256\times 256$	$512$
15		Inception Module #2	Reshape #3	$256\times 256$	Layer Normalization #4	$16\times 16\times 256$	$0$
			Conv2D #7	$16\times 16\times 256$	Reshape #3	$16\times 16\times 96$	$24672$
			Conv2D #8	$16\times 16\times 256$	Reshape #3	$16\times 16\times 16$	$4112$
			Conv2D #9	$16\times 16\times 256$	Reshape #3	$16\times 16\times 64$	16448
			MaxPooling2D #2	$16\times 16\times 256$	Reshape #3	$16\times 16\times 256$	0
			Conv2D #10	$16\times 16\times 96$	Conv2D #7	$16\times 16\times 128$	$110720$
			Conv2D #11	$16\times 16\times 16$	Conv2D #8	$16\times 16\times 32$	$12832$
			Conv2D #12	$16\times 16\times 256$	MaxPooling2D #2	$16\times 16\times 32$	$8224$
			Concatenate #2	$16\times 16\times 64,$ $16\times 16\times 128$ , $16\times 16\times 32$ , $16\times 16\times 32$	Conv2D #9, Conv2D #10, Conv2D #11, Conv2D #12	$16\times 16\times 256$	0
			Reshape #4	$16\times 16\times 256$	Concatenate #2	$256\times 256$	0
16			Add #5	$256\times 256$ , $256\times 256$	Reshape #4, Add #4	$256\times 256$	0
17			GlobalAveragePooling1D #1	$256\times 256$	Add #5	$256$	0
18			Dense #1	$256$	GlobalAveragePooling2D #1	$64$	$16448$
19			Dense #2 (Output Layer)	$64$	Dense #1	4 (for the Maize dataset), 38 (for the Plant-Village dataset)	$262$ , $2470$
Total Weight Parameters						6945574 (for the Maize dataset) 6947782 (for the PlantVillage dataset)

The value written in bold font highlights the layers of Inception module in this table. These layers are the only difference between existing ViT model and our proposed TrIncNet model.