Table 2.

Network architecture: Dimensions of all the intermediate layers of the convolutional neural network. Residual layers contain two feature maps per layer

Input Layer	Input layer dimensions	Filter type	Filter size	Output layer
Input	256 × 256 × 1	Convolutional	3 × 3 × 16	Hidden layer 1
Hidden layer 1	128 × 128 × 16	Residual	3 × 3 × 32	Hidden layer 2/3
Hidden layer 2/3	64 × 64 × 32	Residual	3 × 3 × 64	Hidden layer 4/5
Hidden layer 4/5	32 × 32 × 64	Residual	3 × 3 × 128	Hidden layer 6/7
Hidden layer 6/7	16 × 16 × 128	Residual	3 × 3 × 128	Hidden layer 8/9
hidden layer 8/9	16 × 16 × 128	Residual	3 × 3 × 256	Hidden layer 10/11
Hidden layer 10/11	8 × 8 × 256	Spatial Transform	N/A	Hidden layer 12
Hidden layer12	8 × 8 × 256	Inception	×256	Hidden layer 13
Hidden 13	8 × 8 × 256	Residual	3 × 3 × 512	Hidden layer 14/15
Hidden layer 14/15	4 × 4 × 512	Residual	3 × 3 × 512	Hidden layer 16/17
Hidden layer 16/17	4 × 4 × 512	Residual	3 × 3512	Hidden layer 18/19
Hidden layer 18/19	4 × 4 × 512	Linear	× 16	Hidden layer 20
Hidden layer 20	1 × 16	Linear	16 × 8	Hidden layer 21
Hidden layer 21	1 × 8	Softmax	8 × 3	Classification