Keras: Obtain array of images and labels from flow_from_directory

Question

I'm trying to obtain a confusion matrix using scikit after training my model but since I used flow_from_directory , I don't have access to the data and labels or I don't know a way to do so. Since scikit confusion matrix method is used like:

confusion_matrix(y_true, y_pred)

and flow_from_directory doesn't return the true labels. Is there a way to obtain them from flow_from_directory directly or any other method?

Answer 1

For any generator , you can yield data as with any iterable:

for x, y in something.flow_from_directory(...):
    #do stuff

Be careful to decide when to stop. Generators for training in keras are infinite.