Skip to main content
. 2021 Jul 29;21(15):5125. doi: 10.3390/s21155125

Table 4.

The parameters of MSF-Net.

Layername Type Output
Dimension		 Depth CReLU
Output 		Inception Output Parameters
#1×1-3×3-1×1 Pool
Proj		 #1×1 #3×3 #5×5 #1×1
out
conv1_1 3 × 3 CReLU 224 × 224 × 32 1 NA-16-NA / 480
pool1_1 3 × 3 Max pool. 112 × 112 × 32 0 / /
conv2_1 3 × 3 CReLU 112 × 112 × 64 3 24-24-64 / 13,440
conv2_2 3 × 3 CReLU 112 × 112 × 64 3 24-24-64 13,440
conv2_3 3 × 3 CReLU 112 × 112 × 64 3 24-24-64 13,440
conv3_1 3 × 3 CReLU 56 × 56 × 128 3 48-48-128 / 53,760
conv3_2 3 × 3 CReLU 56 × 56 × 128 3 48-48-128 53,760
conv3_3 3 × 3 CReLU 56 × 56 × 128 3 48-48-128 53,760
conv4_1 Inception 28 × 28 × 256 4 / 32 64 96-128 16-32-32 256 322,560
conv4_2 Inception 28 × 28 × 256 4 32 64 96-128 16-32-32 256 322,560
conv4_3 Inception 28 × 28 × 256 4 32 64 96-128 16-32-32 256 322,560
conv5_1 Inception 14 × 14 × 512 4 / 64 128 128-192 32-96-96 512 1,040,384
conv5_2 Inception 14 × 14 × 512 4 64 128 128-192 32-96-96 512 1,040,384
conv5_3 Inception 14 × 14 × 512 4 64 128 128-192 32-96-96 512 1,040,384
conv5_4 Inception 14 × 14 × 512 4 64 128 128-192 32-96-96 512 1,040,384
conv6_1 Inception 7 × 7 × 1024 4 / 128 256 160-320 32-128-128 1024 3,010,560
conv6-2 Inception 7 × 7 × 1024 4 128 256 160-320 32-128-128 1024 3,010,560
concat Concatenation 7 × 7 × 1920 0 / / /
avg pool 7 × 7 Avg pool. 1 × 1 × 1920 0
linear Inner product 1 × 1 × NClass 1 19,200
Total 47 11,371,616