Figure 8. Training and validation loss plots of model networks with combinations of different numbers of convolutional filter and different sets of training sample. The number of learnable filters used to the model is specified as ‘F-number’ in every sub-caption. TL and VL in legends denote training loss and validation loss. Neighbourhood size is depicted as (N).
By observing the graphs in Figure 7 and Figure 8 some comments can be made: