Fig. 4

Plot of the CNN model’s accuracy and loss on training and validation steps considering the dataset split 80%/20%. The model was trained for 100 epochs. a Accuracy of training and validation considering the binary problem. b Loss of training and validation considering the binary problem. c Accuracy of training and validation considering the multiclass problem. d Loss of training and validation considering the multiclass problem. Each plotted curve is obtained from the history of the Keras model, which calculates both accuracy and loss for each epoch performed by the network. Accuracy is calculated by comparing the predicted class to the actual class. Loss is calculated by the cross entropy value between the predicted class and the actual class