Table 4 Best hyperparameter values for the deep learning model used for spatial resolution and geometric distortion
Hyperparameter | Best Value |
|---|---|
Number of filters (1st convolution layer) | 16 |
Number of filters (2nd convolution layer) | 48 |
Number of filters (3rd convolution layer) | 96 |
Number of filters (4th convolution layer) | 384 |
Number of units in the densely connected layer | 64 |
Learning rate | 0.001 |
Batch size | 64 |
Epochs | 30 |