Table 3 Schematic of the proposed modified Inception V4 CNN. The input size is given in each row, and the output size is the input size of the next row. All convolutions were performed with sigmoid activation and 40% dropout.
Layer | Kernel size/Remarks | Input Size |
|---|---|---|
Conv | 3 × 3/‘valid’ | 101 × 101 × 3 |
Conv | 3 × 3/‘valid’ | 98 × 98 × 32 |
Max Pool | 2 × 2/stride=2 ‘valid’ | 96 × 96 × 64 |
Conv | 3 × 3/stride=2 ‘valid’ | 48 × 48 × 64 |
4 x Inception-A Block | 1 × 1 and 3 × 3/‘same’ | 23 × 23 × 80 |
Reduction-A Block | 1 × 1 and 3 × 3/‘same’ | 23 × 23 × 384 |
7 × Inception Block | 1 × 1, 1 × 7, 7 × 1, and 3 × 3/‘same’ | 11 × 11 × 1024 |
Reduction-B Block | 1 × 1, 1 × 7, 7 × 1, and 3 × 3/‘same’ | 11 × 11×1024 |
3 × Inception-C Block | 1 × 1, 1 × 3, 3 × 1, and 3 × 3/‘same’ | 5 × 5 × 1024 |
Avg. Pool | 5 × 5/‘valid’ | 5 × 5 × 1536 |
Linear | Logits | 1 × 1536 |
Softmax | Classifier | 1 × 2 |
