Good afternoon,
I am new here, so please, if this is not the right location for my question, please give me a pointer…
I use flow_from_dataframe from Keras to prepare image data for processing as Multi-Class-Problem in a very much standard sequential learning setup. My images have 0 to 3 of the three possible classes ‘car’, ‘cat’, ‘human’ attached.
Now this guy
test_generator = test_datagen.flow_from_dataframe(
dataframe=testdata,
directory=dir_dict[‘download’],
target_size=(ydim, xdim),
batch_size=batchsize,
shuffle=False,
classes=classes_list,
class_mode=‘categorical’)
keeps sorting away the images that have none of the classes. They are not being processed. How can I switch of this behaviour?
Thanks in advance
Ludger
I will. Thank you…