preprocess function in Keras doesn't work

Question

In my 2D semantic segmentation task, all pixel values in labels aren't are not 0,1,2, but 0,127,255 for example. So I want to simply add a preprocess function to my ImageDataGenerator of label dataset,

My code:

SEED = 111
batch_size = 2
image_datagen = ImageDataGenerator(
    horizontal_flip=True,
    zca_epsilon=9,
    # fill_mode='nearest',
)
image_generator = image_datagen.flow_from_directory(
    directory="/xxx/images",
    class_mode=None,
    batch_size=batch_size,
    seed=SEED,
)


def preprocessing_function(image):
    # if I have 3 categories, I need to convert 0,10,20 to 0,1,2 for example 
    return image


label_datagen = ImageDataGenerator(
    horizontal_flip=True,
    zca_epsilon=9,
    rescale=1,
    preprocessing_function=preprocessing_function,
    # fill_mode='nearest',
)
label_generator = image_datagen.flow_from_directory(
    directory="/xxx/labels",
    class_mode=None,
    batch_size=batch_size,
    seed=SEED,
)

train_generator = zip(image_generator, label_generator)
print(len(image_generator))
i = 0
for image_batch, label_batch in iter(train_generator):
    print(image_batch.shape, label_batch.shape) # (2, 256, 256, 3) (2, 256, 256, 3)
    print(image_batch.dtype, label_batch.dtype) # float32 float32
    i += 1
    if i == 5:
        break

But it seems that my

preprocessing_function(image)

has no effect on my label data.

Am I using the preprocess function in the right way? How can I repair this?

Answer 1

I find the solution:

If I pass a preprocessing function to label data's ImageDataGenerator(), I need to use:

label_batch = label_datagen.standardize(label_batch)

on each of my label batch to activate the preprocessing function.