Skip to content

reflection_padding_2d

ReflectionPadding2D

Bases: Layer

A layer for performing Reflection Padding on 2D arrays.

This class is intentionally not @traceable (models and layers are handled by a different process).

This layer assumes that you are using the a tensor shaped like (Batch, Height, Width, Channels). The implementation here is borrowed from https://stackoverflow.com/questions/50677544/reflection-padding-conv2d.

x = tf.reshape(tf.convert_to_tensor(list(range(9))), (1,3,3,1))  # ~ [[0, 1, 2], [3, 4, 5], [6, 7, 8]]
m = fe.layers.tensorflow.ReflectionPadding2D((1, 1))
y = m(x)  # ~ [[4, 3, 4, 5, 4], [1, 0, 1, 2, 1], [4, 3, 4, 5, 4], [7, 6, 7, 8, 7], [4, 3, 4, 5, 4]]
m = fe.layers.tensorflow.ReflectionPadding2D((1, 0))
y = m(x)  # ~ [[1, 0, 1, 2, 1], [4, 3, 4, 5, 4], [7, 6, 7, 8, 7]]

Parameters:

Name Type Description Default
padding Tuple[int, int]

padding size (Width, Height). The padding size must be less than the size of the corresponding dimension in the input tensor.

 (1, 1)
Source code in fastestimator/fastestimator/layers/tensorflow/reflection_padding_2d.py 
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
class ReflectionPadding2D(layers.Layer):
    """A layer for performing Reflection Padding on 2D arrays.

    This class is intentionally not @traceable (models and layers are handled by a different process).

    This layer assumes that you are using the a tensor shaped like (Batch, Height, Width, Channels).
    The implementation here is borrowed from https://stackoverflow.com/questions/50677544/reflection-padding-conv2d.

    ```python
    x = tf.reshape(tf.convert_to_tensor(list(range(9))), (1,3,3,1))  # ~ [[0, 1, 2], [3, 4, 5], [6, 7, 8]]
    m = fe.layers.tensorflow.ReflectionPadding2D((1, 1))
    y = m(x)  # ~ [[4, 3, 4, 5, 4], [1, 0, 1, 2, 1], [4, 3, 4, 5, 4], [7, 6, 7, 8, 7], [4, 3, 4, 5, 4]]
    m = fe.layers.tensorflow.ReflectionPadding2D((1, 0))
    y = m(x)  # ~ [[1, 0, 1, 2, 1], [4, 3, 4, 5, 4], [7, 6, 7, 8, 7]]
    ```

    Args:
        padding: padding size (Width, Height). The padding size must be less than the size of the corresponding
            dimension in the input tensor.
    """
    def __init__(self, padding: Tuple[int, int] = (1, 1)) -> None:
        super().__init__()
        self.padding = tuple(padding)
        self.input_spec = [layers.InputSpec(ndim=4)]

    def get_config(self) -> Dict[str, Any]:
        return {'padding': self.padding}

    def compute_output_shape(self, s: Tuple[int, int, int, int]) -> Tuple[int, int, int, int]:
        """If you are using "channels_last" configuration"""
        return s[0], s[1] + 2 * self.padding[0], s[2] + 2 * self.padding[1], s[3]

    def call(self, x: tf.Tensor) -> tf.Tensor:
        w_pad, h_pad = self.padding
        return tf.pad(x, [[0, 0], [h_pad, h_pad], [w_pad, w_pad], [0, 0]], 'REFLECT')

compute_output_shape

If you are using "channels_last" configuration

Source code in fastestimator/fastestimator/layers/tensorflow/reflection_padding_2d.py 
49
50
51
def compute_output_shape(self, s: Tuple[int, int, int, int]) -> Tuple[int, int, int, int]:
    """If you are using "channels_last" configuration"""
    return s[0], s[1] + 2 * self.padding[0], s[2] + 2 * self.padding[1], s[3]