ImageDataGenerator without classes

[TheStrange-007]

Body

How can I use the ImageDataGenerator in Keras to handle images without organizing them into separate classes within folders? For example, if I have a collection of images for training that aren't categorized into separate folders for classes.

I want to utilize the grayscale version of the respective image as the input (X) and its RGB counterpart as the output (y).

Guidelines

• I have read and understood this category's guidelines before making this post.

[faridnec]

Hey, You can perhaps use the flow_from_directory method of ImageDataGenerator in Keras to handle images even if they are not organized into separate classes within folders. Instead of organizing images into separate class folders, you can organize them all into a single directory and specify the class_mode parameter as None. This tells Keras not to expect subdirectories for classes.

train_generator = datagen.flow_from_directory(
        train_dir,
        target_size=(image_height, image_width),
        batch_size=batch_size,
        class_mode=None,  # Set class_mode to None since there are no classes
...................................................