Skip to content

cropping_2d

Cropping2D

Bases: Module

A layer for cropping along height and width dimensions.

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

x = torch.tensor(list(range(100))).view((1,1,10,10))
m = fe.layers.pytorch.Cropping2D(3)
y = m.forward(x)  # [[[[33, 34, 35, 36], [43, 44, 45, 46], [53, 54, 55, 56], [63, 64, 65, 66]]]]
m = fe.layers.pytorch.Cropping2D((3, 4))
y = m.forward(x)  # [[[[34, 35], [44, 45], [54, 55], [64, 65]]]]
m = fe.layers.pytorch.Cropping2D(((1, 4), 4))
y = m.forward(x)  # [[[[14, 15], [24, 25], [34, 35], [44, 45], [54, 55]]]]

Parameters:

Name Type Description Default
cropping Union[int, Tuple[Union[int, Tuple[int, int]], Union[int, Tuple[int, int]]]]

Height and width cropping parameters. If a single int 'n' is specified, then the width and height of the input will both be reduced by '2n', with 'n' coming off of each side of the input. If a tuple ('h', 'w') is provided, then the height and width of the input will be reduced by '2h' and '2w' respectively, with 'h' and 'w' coming off of each side of the input. If a tuple like (('h1', 'h2'), ('w1', 'w2')) is provided, then 'h1' will be removed from the top, 'h2' from the bottom, 'w1' from the left, and 'w2' from the right (assuming the top left corner as the 0,0 origin).

 0

Raises:

Type Description
ValueError

If cropping has an unacceptable data type.
Source code in fastestimator/fastestimator/layers/pytorch/cropping_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
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
class Cropping2D(nn.Module):
    """A layer for cropping along height and width dimensions.

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

    ```python
    x = torch.tensor(list(range(100))).view((1,1,10,10))
    m = fe.layers.pytorch.Cropping2D(3)
    y = m.forward(x)  # [[[[33, 34, 35, 36], [43, 44, 45, 46], [53, 54, 55, 56], [63, 64, 65, 66]]]]
    m = fe.layers.pytorch.Cropping2D((3, 4))
    y = m.forward(x)  # [[[[34, 35], [44, 45], [54, 55], [64, 65]]]]
    m = fe.layers.pytorch.Cropping2D(((1, 4), 4))
    y = m.forward(x)  # [[[[14, 15], [24, 25], [34, 35], [44, 45], [54, 55]]]]
    ```

    Args:
        cropping: Height and width cropping parameters. If a single int 'n' is specified, then the width and height of
            the input will both be reduced by '2n', with 'n' coming off of each side of the input. If a tuple ('h', 'w')
            is provided, then the height and width of the input will be reduced by '2h' and '2w' respectively, with 'h'
            and 'w' coming off of each side of the input. If a tuple like (('h1', 'h2'), ('w1', 'w2')) is provided, then
            'h1' will be removed from the top, 'h2' from the bottom, 'w1' from the left, and 'w2' from the right
            (assuming the top left corner as the 0,0 origin).

    Raises:
        ValueError: If `cropping` has an unacceptable data type.
    """
    def __init__(self, cropping: Union[int, Tuple[Union[int, Tuple[int, int]], Union[int, Tuple[int,
                                                                                                int]]]] = 0) -> None:
        super().__init__()

        if isinstance(cropping, int):
            self.cropping = ((cropping, cropping), (cropping, cropping))
        elif hasattr(cropping, '__len__'):
            if len(cropping) != 2:
                raise ValueError(f"'cropping' should have two elements, but found {len(cropping)}")
            if isinstance(cropping[0], int):
                height_cropping = (cropping[0], cropping[0])
            elif hasattr(cropping[0], '__len__') and len(cropping[0]) == 2:
                height_cropping = (cropping[0][0], cropping[0][1])
            else:
                raise ValueError(f"'cropping' height should be an int or tuple of ints, but found {cropping[0]}")
            if isinstance(cropping[1], int):
                width_cropping = (cropping[1], cropping[1])
            elif hasattr(cropping[1], '__len__') and len(cropping[1]) == 2:
                width_cropping = (cropping[1][0], cropping[1][1])
            else:
                raise ValueError(f"'cropping' width should be an int or tuple of ints, but found {cropping[1]}")
            self.cropping = (height_cropping, width_cropping)
        else:
            raise ValueError(
                "cropping` should be either an int, a tuple of 2 ints or a tuple of two tuple of 2 ints. Found: " +
                str(cropping))

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        return x[:, :, self.cropping[0][0]:-self.cropping[0][1], self.cropping[1][0]:-self.cropping[1][1]]