Search code examples
pythontensorflowkerasdata-augmentation

Apply a transformation model (data augmentation) in images in Tensorflow


I am a newbie in some sequential models in Tensorflow with Python. I have a transformation sequential model like the one below. It applies randomly to a given image input some operations with random parameters.

import tensorflow as tf
from tensorflow.keras import layers

data_transformation = tf.keras.Sequential(
    [
        layers.Lambda(lambda x: my_random_brightness(x, 1, 20)))
        layers.GaussianNoise(stddev=tf.random.uniform(shape=(), minval=0, maxval=1)),
        layers.experimental.preprocessing.RandomRotation(
            factor=0.01,
            fill_mode="reflect",
            interpolation="bilinear",
            seed=None,
            name=None,
            fill_value=0.0,
        ),
        layers.experimental.preprocessing.RandomZoom(
            height_factor=(0.1, 0.2),
            width_factor=(0.1, 0.2),
            fill_mode="reflect",
            interpolation="bilinear",
            seed=None,
            name=None,
            fill_value=0.0,
        ),
    ]
)

There is also a lambda function in this model that define as below

def my_random_brightness(
    image_to_be_transformed, brightness_factor_min, brightness_factor_max
):

    # build the brightness factor
    selected_brightness_factor = tf.random.uniform(
        (), minval=brightness_factor_min, maxval=brightness_factor_max
    )

    c0 = image_to_be_transformed[:, :, :, 0] + selected_brightness_factor
    c1 = image_to_be_transformed[:, :, :, 1] + selected_brightness_factor
    c2 = image_to_be_transformed[:, :, :, 2] + selected_brightness_factor

    image_to_be_transformed = tf.concat(
        [c0[..., tf.newaxis], image_to_be_transformed[:, :, :, 1:]], axis=-1
    )

    image_to_be_transformed = tf.concat(
        [
            image_to_be_transformed[:, :, :, 0][..., tf.newaxis],
            c1[..., tf.newaxis],
            image_to_be_transformed[:, :, :, 2][..., tf.newaxis],
        ],
        axis=-1,
    )

    image_to_be_transformed = tf.concat(
        [image_to_be_transformed[:, :, :, :2], c2[..., tf.newaxis]], axis=-1
    )

    return image_to_be_transformed

Just now suppose that I would like to apply such a model to output such random operations in one batch containing just one image and I would like to see and save the result. How is that possible to do that? is there any predict() or flow() like function to output such a result?

EDIT: I tried result=data_transformation(image) and I have the following error:

tensorflow.python.framework.errors_impl.InvalidArgumentError: Index out of range using input dim 3; input has only 3 dims [Op:StridedSlice] name: sequential/lambda/strided_slice/


Solution

  • Apart from the correctness of the brightness processing layer (above), it's coded to take a batch of images and not a single image. That's the reason it gives the expected error. You should add a batch axis while passing a single image in this case. It should work.

    result=data_transformation(image[None, ...])
    

    Also, in custom layer implementation, try always to adopt subclassing way.

    class MyCustomBrightNess(layers.Layer):
        def __init__(self, pbrightness_factor_min, brightness_factor_max, **kwargs):
            super().__init__(**kwargs)
            self.brightness_factor_max = brightness_factor_max
            self.pbrightness_factor_min = pbrightness_factor_min
            
        def call(self, inputs):
             # build the brightness factor
          selected_brightness_factor = tf.random.uniform(
             (), minval=self.brightness_factor_min, maxval=self.brightness_factor_max
          )
    
          c0 = inputs[:, :, :, 0] + selected_brightness_factor
          c1 = inputs[:, :, :, 1] + selected_brightness_factor
          c2 = inputs[:, :, :, 2] + selected_brightness_factor
    
          inputs = tf.concat(
             [c0[..., tf.newaxis], inputs[:, :, :, 1:]], axis=-1
          )
    
          inputs = tf.concat(
             [
                   inputs[:, :, :, 0][..., tf.newaxis],
                   c1[..., tf.newaxis],
                   inputs[:, :, :, 2][..., tf.newaxis],
             ],
             axis=-1,
          )
    
          inputs = tf.concat(
             [inputs[:, :, :, :2], c2[..., tf.newaxis]], axis=-1
          )
    
          return inputs
            
        def get_config(self):
            config = {
                'pbrightness_factor_min': self.pbrightness_factor_min,
                'brightness_factor_max': self.brightness_factor_max
            }
            base_config = super(MyCustomBrightNess, self).get_config()
            return dict(list(base_config.items()) + list(config.items()))
    

    About the correctness of this implementation, I didn't check rigorously. I would suggest using random_brightness or adjust_brightness from the official implementation. Or if you're using tensorflow2.9, say hello to the new KerasCV, there we can find RandomBrightness layers.