Table 3. CNN architecture.

Network architecture	Output
Input layer (Image dimensions: 40 × 862 × 1)
Conv2D (16F, K2, ReLu) - MaxPooling2D (P2) - Dr 0.2	19 × 430 × 16
Conv2D (32F, K2, ReLu) - MaxPooling2D (P2) - Dr 0.2	9 × 214 × 32
Conv2D (64F, K2, ReLu) - MaxPooling2D (P2) - Dr 0.2	4 × 106 × 64
Conv2D (128F, K2, ReLu) - MaxPooling2D (P2) - Dr 0.2	1 × 52 × 128
GlobalAverage2D	128
Dense	2