Table 3.

Different layers of the proposed Architecture for the automatic detection of pneumonia

S. No	Layer (type)	Stride	Filter Shape	Output Shape
1	Input Layer	-	-	(Batch_size, 512, 512, 3)
2	Conv2D	2	(3, 3, 3, 32)	(Batch_size, 256, 256, 32)
3	DepthwiseConv2D	1	(3, 3, 32)	( Batch_size, 256, 256, 32)
4	Conv2D	1	(1, 1, 32, 64)	( Batch_size, 256, 256, 64)
5	DepthwiseConv2D	2	(3, 3, 64)	( Batch_size, 128, 128, 64)
6	Conv2D	1	(1, 1, 64, 128)	( Batch_size, 128, 128, 128)
7	DepthwiseConv2D	1	(3, 3, 128)	( Batch_size, 128, 128, 128)
8	Conv2D	1	(1, 1, 128, 128)	( Batch_size, 128, 128, 128)
9	DepthwiseConv2D	2	(3, 3, 128)	( Batch_size, 64, 64, 128)
10	Conv2D	1	(1, 1, 128, 256)	( Batch_size, 64, 64, 256)
11	DepthwiseConv2D	1	(3, 3, 256)	( Batch_size, 64, 64, 256)
12	Conv2D	1	(1, 1, 256, 256)	( Batch_size, 64, 64, 256)
13	DepthwiseConv2D	2	(3, 3, 256)	( Batch_size, 32, 32, 256)
14	Conv2D	1	(1, 1, 256, 512)	( Batch_size, 32, 32, 512)
15	DepthwiseConv2D	1	(3, 3, 512)	( Batch_size, 32, 32, 512)
16	Conv2D	1	(1, 1, 512, 512)	( Batch_size, 32, 32, 512)
17	DepthwiseConv2D	1	(3, 3, 512)	( Batch_size, 32, 32, 512)
18	Conv2D	1	(1, 1, 512, 512)	( Batch_size, 32, 32, 512)
19	DepthwiseConv2D	1	(3, 3, 512)	( Batch_size, 32, 32, 512)
20	Conv2D	1	(1, 1, 512, 512)	( Batch_size, 32, 32, 512)
21	DepthwiseConv2D	1	(3, 3, 512)	( Batch_size, 32, 32, 512)
22	Conv2D	1	(1, 1, 512, 512)	( Batch_size, 32, 32, 512)
23	DepthwiseConv2D	1	(3, 3, 512)	( Batch_size, 32, 32, 512)
24	Conv2D	1	(1, 1, 512, 512)	( Batch_size, 32, 32, 512)
25	DepthwiseConv2D	2	(3, 3, 512)	( Batch_size, 16, 16, 512)
26	Conv2D	1	(1, 1, 512, 1024)	( Batch_size, 16, 16, 1024)
27	DepthwiseConv2D	2	(3, 3, 1024)	( Batch_size, 16, 16, 1024)
28	Conv2D	1	(1, 1, 1024, 1024)	( Batch_size, 16, 16, 1024)
29	GlobalAveragePooling2D	1	Pool (16, 16)	( Batch_size, 1024)
30	Sigmoid Classifier	-	Classifier	( Batch_size, 1)
Total params: 3,229,889 Trainable params: 3,208,001 Non-trainable params: 21,888