Table 3 A layer-by-layer comprehensive outline of the maizenet model.
From: Enhanced residual-attention deep neural network for disease classification in maize leaf images
Layer Name | Model Layer | Filter Size | Strides | Output Dimension | Layerwise-Parameters |
|---|---|---|---|---|---|
Input_1 | Input | - | - | 224 × 224 × 3 | 0 |
Cblr_1 | Conv2D_BNz_LeakyRelu | 3 × 3 | 1 | 224 × 224 × 16 | 432 + 64 + 0 |
Cblr_2 | Conv2D _BNz_LeakyRelu | 3 × 3 | 1 | 224 × 224 × 32 | 4640 + 128 + 0 |
gap_1 | GAP-2D | - | - | 32 | 0 |
reshape | Reshape | - | - | 1 × 1 × 32 | 0 |
dense | Dense | - | - | 1 × 1 × 2 | 64 |
dense_1 | Dense | - | - | 1 × 1 × 32 | 64 |
conv2d_1 | Conv2D | 1 × 1 | 1 | 224 × 224 × 1 | 32 |
multiply | Multiply | - | - | 224 × 224 × 32 | 0 |
multiply_1 | Multiply | - | - | 224 × 224 × 32 | 0 |
add_1 | Add | - | - | 224 × 224 × 32 | 0 |
Blrc_1 | BNz_LeakyRelu_Conv2D | 3 × 3 | 2 | 112 × 112 × 32 | 128 + 0 + 9248 |
Blrc_2 | BNz_LeakyRelu_Conv2D | 3 × 3 | 1 | 112 × 112 × 32 | 128 + 0 + 1056 |
conv2d_2 | Conv2D | 1 × 1 | 1 | 112 × 112 × 32 | 9248 |
add_2 | Add | - | - | 112 × 112 × 32 | 0 |
Cblr_3 | Conv2D _BNz_LeakyRelu | 3 × 3 | 1 | 112 × 112 × 64 | 18,496 + 256 + 0 |
gap_2 | GAP-2D | - | - | 64 | 0 |
Reshape_1 | Reshape | - | - | 1 × 1 × 64 | 0 |
dense_2 | Dense | - | - | 1 × 1 × 4 | 256 |
dense_3 | Dense | - | - | 1 × 1 × 64 | 256 |
conv2d_3 | Conv2D | 1 × 1 | 112 × 112 × 1 | 64 | |
multiply_2 | Multiply | - | 1 | 112 × 112 × 64 | 0 |
multiply_3 | Multiply | - | 1 | 112 × 112 × 64 | 0 |
add_3 | Add | - | - | 112 × 112 × 64 | 0 |
Blrc_3 | BNz_LeakyRelu_Conv2D | 3 × 3 | 2 | 56 × 56 × 64 | 256 + 0 + 36,928 |
Blrc_4 | BNz_LeakyRelu_Conv2D | 3 × 3 | 1 | 56 × 56 × 64 | 256 + 0 + 4160 |
conv2d_4 | Conv2D | 1 × 1 | 1 | 56 × 56 × 64 | 36,928 |
add_4 | Add | - | - | 56 × 56 × 64 | 0 |
Cblr_4 | Conv2D _BNz_LeakyRelu | 3 × 3 | 1 | 56 × 56 × 128 | 73,856 + 512 + 0 |
gap_3 | GAP-2D | - | - | 128 | 0 |
reshape | Reshape | - | - | 1 × 1 × 128 | 0 |
dense_4 | Dense | - | - | 1 × 1 × 8 | 1024 |
dense_5 | Dense | - | - | 1 × 1 × 128 | 1024 |
conv2d_5 | Conv2D | 1 × 1 | 1 | 56 × 56 × 1 | 128 |
multiply_4 | Multiply | - | - | 56 × 56 × 128 | 0 |
multiply_5 | Multiply | - | - | 56 × 56 × 128 | 0 |
add_5 | Add | - | - | 56 × 56 × 128 | 0 |
Blrc_5 | BNz_LeakyRelu_Conv2D | 3 × 3 | 2 | 28 × 28 × 128 | 512 + 0 + 147,584 |
Blrc_6 | BN_LeakyRelu_Conv2D | 3 × 3 | 1 | 28 × 28 × 128 | 512 + 0 + 16,512 |
conv2d_6 | Conv2D | 1 × 1 | 1 | 28 × 28 × 128 | 147,584 |
add_6 | Add | - | - | 28 × 28 × 128 | 0 |
Cblr_5 | Conv2D _BNz_LeakyRelu | 3 × 3 | 1 | 28 × 28 × 256 | 295,168 + 1024 + 0 |
gap_4 | GAP-2D | - | - | 256 | 0 |
reshape | Reshape | - | - | 1 × 1 × 256 | 0 |
dense_6 | Dense | - | - | 1 × 1 × 16 | 4096 |
dense_7 | Dense | - | - | 1 × 1 × 64 | 4096 |
conv2d_7 | Conv2D | 1 × 1 | 1 | 28 × 28 × 1 | 256 |
multiply_6 | Multiply | - | - | 28 × 28 × 256 | 0 |
multiply_7 | Multiply | - | - | 28 × 28 × 256 | 0 |
add_7 | Add | - | - | 28 × 28 × 256 | 0 |
Blrc_7 | BNz_LeakyRelu_Conv2D | 3 × 3 | 2 | 14 × 14 × 256 | 1024 + 0 + 590,080 |
Blrc_8 | BNz_LeakyRelu_Conv2D | 3 × 3 | 1 | 14 × 14 × 256 | 1024 + 0 + 65,792 |
conv2d_8 | Conv2D | 1 × 1 | 1 | 14 × 14 × 256 | 590,080 |
add_8 | Add | - | 1 | 14 × 14 × 256 | 0 |
gap_5 | GAP-2D | - | - | 256 | 0 |
dense_8 | Dense | - | - | 128 | 32,896 |
dropout_1 | Dropout | - | - | 128 | 0 |
dense_9 | Dense | - | - | 64 | 8256 |
dropout_2 | Dropout | - | - | 64 | 0 |
dense_10 | Dense | - | - | 4 | 260 |
Total parameters | 21,06,388 | ||||
Trainable parameters | 21,03,476 | ||||
Non-trainable parameters | 2,912 | ||||
