Skip to content

SynthShapes.utils

MinMaxScaler 

MinMaxScaler(lower_bound=0.01, upper_bound=0.99)

Bases: Module

A torch.nn.Module subclass that scales an input tensor to a specified range.

PARAMETER DESCRIPTION
lower_bound

The lower bound of the scaling interval. Default is 0.1.

TYPE: float DEFAULT: 0.01

upper_bound

The upper bound of the scaling interval. Default is 0.9.

TYPE: float DEFAULT: 0.99

Source code in SynthShapes/utils.py 
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
def __init__(self, lower_bound: float = 0.01, upper_bound: float = 0.99):
    """
    A torch.nn.Module subclass that scales an input tensor to a specified
    range.

    Parameters
    ----------
    lower_bound : float
        The lower bound of the scaling interval. Default is 0.1.
    upper_bound : float
        The upper bound of the scaling interval. Default is 0.9.
    """
    super(MinMaxScaler, self).__init__()
    self.lower_bound = lower_bound
    self.upper_bound = upper_bound

forward 

forward(X)

Forward pass to scale the input tensor X to the interval [lower_bound, upper_bound].

PARAMETER DESCRIPTION
X

Input tensor with arbitrary floating point values.

TYPE: Tensor

RETURNS DESCRIPTION
Tensor

Scaled tensor with values in the interval [lower_bound, upper_bound].
Source code in SynthShapes/utils.py 
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
def forward(self, X: torch.Tensor) -> torch.Tensor:
    """
    Forward pass to scale the input tensor X to the interval
    [lower_bound, upper_bound].

    Parameters
    ----------
    X : torch.Tensor
        Input tensor with arbitrary floating point values.

    Returns
    -------
    torch.Tensor
        Scaled tensor with values in the interval
        [lower_bound, upper_bound].
    """
    X_min = X.min()
    X_max = X.max()

    # Prevent division by zero if X_min == X_max
    if X_min == X_max:
        return torch.full_like(X, self.lower_bound)

    # Scale the tensor to the range [lower_bound, upper_bound]
    X_scaled = self.lower_bound + (X - X_min) * (
        self.upper_bound - self.lower_bound) / (X_max - X_min)

    return X_scaled

ensure_5d_tensor 

ensure_5d_tensor(volume)

Ensures the input tensor has 5 dimensions [batch_size, channels, depth, height, width].

PARAMETER DESCRIPTION
volume

Input tensor representing the volume.

TYPE: Tensor

RETURNS DESCRIPTION
Tensor

Tensor with 5 dimensions.
Source code in SynthShapes/utils.py 
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
def ensure_5d_tensor(volume: torch.Tensor) -> torch.Tensor:
    """
    Ensures the input tensor has 5 dimensions [batch_size, channels, depth,
    height, width].

    Parameters
    ----------
    volume : torch.Tensor
        Input tensor representing the volume.

    Returns
    -------
    torch.Tensor
        Tensor with 5 dimensions.
    """
    if volume.dim() == 3:  # (D, H, W)
        # Add batch and channel dimensions
        volume = volume.unsqueeze(0).unsqueeze(0)
    elif volume.dim() == 4:  # (C, D, H, W)
        # Add batch dimension
        volume = volume.unsqueeze(0)

    return volume