Dice Score¶

Module Interface¶

class torchmetrics.segmentation.DiceScore(num_classes, include_background=True, average='micro', input_format='one-hot', **kwargs)[source]

Compute Dice Score.

The metric can be used to evaluate the performance of image segmentation models. The Dice Score is defined as:

..math::: DS = frac{2 sum_{i=1}^{N} t_i p_i}{sum_{i=1}^{N} t_i + sum_{i=1}^{N} p_i}

where \(N\) is the number of classes, \(t_i\) is the target tensor, and \(p_i\) is the prediction tensor. In general the Dice Score can be interpreted as the overlap between the prediction and target tensors divided by the total number of elements in the tensors.

As input to forward and update the metric accepts the following input:

preds (Tensor): An one-hot boolean tensor of shape (N, C, ...) with N being the number of samples and C the number of classes. Alternatively, an integer tensor of shape (N, ...) can be provided, where the integer values correspond to the class index. The input type can be controlled with the input_format argument.

target (Tensor): An one-hot boolean tensor of shape (N, C, ...) with N being the number of samples and C the number of classes. Alternatively, an integer tensor of shape (N, ...) can be provided, where the integer values correspond to the class index. The input type can be controlled with the input_format argument.

As output to forward and compute the metric returns the following output:

gds (Tensor): The dice score. If average is set to None or "none" the output will be a tensor of shape (C,) with the dice score for each class. If average is set to "micro", "macro", or "weighted" the output will be a scalar tensor. The score is an average over all samples.

Parameters:

num_classes¶ (int) – The number of classes in the segmentation problem.
include_background¶ (bool) – Whether to include the background class in the computation.
average¶ (Optional[Literal['micro', 'macro', 'weighted', 'none']]) – The method to average the dice score. Options are "micro", "macro", "weighted", "none" or None. This determines how to average the dice score across different classes.
input_format¶ (Literal['one-hot', 'index']) – What kind of input the function receives. Choose between "one-hot" for one-hot encoded tensors or "index" for index tensors
kwargs¶ (Any) – Additional keyword arguments, see Advanced metric settings for more info.

Raises:

ValueError – If num_classes is not a positive integer
ValueError – If include_background is not a boolean
ValueError – If average is not one of "micro", "macro", "weighted", "none" or None
ValueError – If input_format is not one of "one-hot" or "index"

Example

>>> from torch import randint
>>> from torchmetrics.segmentation import DiceScore
>>> preds = randint(0, 2, (4, 5, 16, 16))  # 4 samples, 5 classes, 16x16 prediction
>>> target = randint(0, 2, (4, 5, 16, 16))  # 4 samples, 5 classes, 16x16 target
>>> dice_score = DiceScore(num_classes=5, average="micro")
>>> dice_score(preds, target)
tensor(0.4941)
>>> dice_score = DiceScore(num_classes=5, average="none")
>>> dice_score(preds, target)
tensor([0.4860, 0.4999, 0.5014, 0.4885, 0.4915])

compute()[source]

Computes the Dice Score.

Return type:: Tensor

plot(val=None, ax=None)[source]

Plot a single or multiple values from the metric.

Parameters:

val¶ (Union[Tensor, Sequence[Tensor], None]) – Either a single result from calling metric.forward or metric.compute or a list of these results. If no value is provided, will automatically call metric.compute and plot that result.
ax¶ (Optional[Axes]) – An matplotlib axis object. If provided will add plot to that axis

Return type:

tuple[Figure, Union[Axes, ndarray]]

Returns:

Figure and Axes object

Raises:

ModuleNotFoundError – If matplotlib is not installed

>>> # Example plotting a single value
>>> import torch
>>> from torchmetrics.segmentation import DiceScore
>>> metric = DiceScore(num_classes=3)
>>> metric.update(torch.randint(0, 2, (10, 3, 128, 128)), torch.randint(0, 2, (10, 3, 128, 128)))
>>> fig_, ax_ = metric.plot()

>>> # Example plotting multiple values
>>> import torch
>>> from torchmetrics.segmentation import DiceScore
>>> metric = DiceScore(num_classes=3)
>>> values = [ ]
>>> for _ in range(10):
...     values.append(
...        metric(torch.randint(0, 2, (10, 3, 128, 128)), torch.randint(0, 2, (10, 3, 128, 128)))
...     )
>>> fig_, ax_ = metric.plot(values)

update(preds, target)[source]

Update the state with new data.

Return type:: None

Functional Interface¶

torchmetrics.functional.segmentation.dice_score(preds, target, num_classes, include_background=True, average='micro', input_format='one-hot')[source]

Compute the Dice score for semantic segmentation.

Parameters:

preds¶ (Tensor) – Predictions from model
target¶ (Tensor) – Ground truth values
num_classes¶ (int) – Number of classes
include_background¶ (bool) – Whether to include the background class in the computation
average¶ (Optional[Literal['micro', 'macro', 'weighted', 'none']]) – The method to average the dice score. Options are "micro", "macro", "weighted", "none" or None. This determines how to average the dice score across different classes.
input_format¶ (Literal['one-hot', 'index']) – What kind of input the function receives. Choose between "one-hot" for one-hot encoded tensors or "index" for index tensors

Return type:

Tensor

Returns:

The Dice score.

Example (with one-hot encoded tensors):

>>> from torch import randint
>>> from torchmetrics.functional.segmentation import dice_score
>>> preds = randint(0, 2, (4, 5, 16, 16))  # 4 samples, 5 classes, 16x16 prediction
>>> target = randint(0, 2, (4, 5, 16, 16))  # 4 samples, 5 classes, 16x16 target
>>> # dice score micro averaged over all classes
>>> dice_score(preds, target, num_classes=5, average="micro")
tensor([0.4842, 0.4968, 0.5053, 0.4902])
>>> # dice score per sample and class
>>> dice_score(preds, target, num_classes=5, average="none")
tensor([[0.4724, 0.5185, 0.4710, 0.5062, 0.4500],
        [0.4571, 0.4980, 0.5191, 0.4380, 0.5649],
        [0.5428, 0.4904, 0.5358, 0.4830, 0.4724],
        [0.4715, 0.4925, 0.4797, 0.5267, 0.4788]])

Example (with index tensors):

>>> from torch import randint
>>> from torchmetrics.functional.segmentation import dice_score
>>> preds = randint(0, 5, (4, 16, 16))  # 4 samples, 5 classes, 16x16 prediction
>>> target = randint(0, 5, (4, 16, 16))  # 4 samples, 5 classes, 16x16 target
>>> # dice score micro averaged over all classes
>>> dice_score(preds, target, num_classes=5, average="micro", input_format="index")
tensor([0.2031, 0.1914, 0.2500, 0.2266])
>>> # dice score per sample and class
>>> dice_score(preds, target, num_classes=5, average="none", input_format="index")
tensor([[0.1714, 0.2500, 0.1304, 0.2524, 0.2069],
        [0.1837, 0.2162, 0.0962, 0.2692, 0.1895],
        [0.3866, 0.1348, 0.2526, 0.2301, 0.2083],
        [0.1978, 0.2804, 0.1714, 0.1915, 0.2783]])