!11195 develop dice loss

From: @lijiaqi0612
Reviewed-by: 
Signed-off-by:

mindspore/nn/loss/__init__.py

@@ -21,8 +21,8 @@ It shows how well the model works on a dataset and the optimization target which
 
 from .loss import L1Loss, MSELoss, SmoothL1Loss, \
     SoftmaxCrossEntropyWithLogits, BCELoss, CosineEmbeddingLoss, \
-    SampledSoftmaxLoss
+    SampledSoftmaxLoss, DiceLoss
 
 __all__ = ['L1Loss', 'MSELoss', 'SmoothL1Loss',
            'SoftmaxCrossEntropyWithLogits', 'BCELoss',
-           'CosineEmbeddingLoss', 'SampledSoftmaxLoss']
+           'CosineEmbeddingLoss', 'SampledSoftmaxLoss', 'DiceLoss']

mindspore/nn/loss/loss.py

@@ -297,6 +297,67 @@ def _check_label_dtype(labels_dtype, cls_name):
     validator.check_type_name("labels", labels_dtype, [mstype.int32, mstype.int64], cls_name)
 
 
+class DiceLoss(_Loss):
+    r"""
+    The Dice coefficient is a set similarity loss. It is used to calculate the similarity between two samples. The
+    value of the Dice coefficient is 1 when the segmentation result is the best and 0 when the segmentation result
+    is the worst. The Dice coefficient indicates the ratio of the area between two objects to the total area.
+    The function is shown as follows:
+
+    .. math::
+        dice = 1 - \frac{2 * (pred \bigcap true)}{pred \bigcup true}
+
+    Args:
+        smooth (float): A term added to the denominator to improve numerical stability. Should be greater than 0.
+                        Default: 1e-5.
+        threshold (float): A threshold, which is used to compare with the input tensor. Default: 0.5.
+
+    Inputs:
+        - **y_pred** (Tensor) - Tensor of shape (N, C).
+        - **y** (Tensor) - Tensor of shape (N, C).
+
+    Outputs:
+        Tensor, a tensor of shape with the per-example sampled Dice losses.
+
+    Supported Platforms:
+        ``Ascend``
+
+    Examples:
+        >>> loss = nn.Diceloss(smooth=1e-5, threshold=0.5)
+        >>> y_pred = Tensor(np.array([[0.2, 0.5], [0.3, 0.1], [0.9, 0.6]]), mstype.float32)
+        >>> y = Tensor(np.array([[0, 1], [1, 0], [0, 1]]), mstype.float32)
+        >>> output = loss(y_pred, y)
+        >>> print(output)
+        [0.77777076]
+    """
+    def __init__(self, smooth=1e-5, threshold=0.5):
+        super(DiceLoss, self).__init__()
+        self.smooth = validator.check_positive_float(smooth, "smooth")
+        self.threshold = validator.check_value_type("threshold", threshold, [float])
+        self.reshape = P.Reshape()
+
+    def construct(self, logits, label):
+        _check_shape(logits.shape, label.shape)
+        logits = self.cast((logits > self.threshold), mstype.float32)
+        label = self.cast(label, mstype.float32)
+        dim = label.shape
+        pred_flat = self.reshape(logits, (dim[0], -1))
+        true_flat = self.reshape(label, (dim[0], -1))
+
+        intersection = self.reduce_sum((pred_flat * true_flat), 1)
+        unionset = self.reduce_sum(pred_flat, 1) + self.reduce_sum(true_flat, 1)
+
+        dice = (2 * intersection + self.smooth) / (unionset + self.smooth)
+        dice_loss = 1 - self.reduce_sum(dice) / dim[0]
+
+        return dice_loss
+
+
+@constexpr
+def _check_shape(logits_shape, label_shape):
+    validator.check('logits_shape', logits_shape, 'label_shape', label_shape)
+
+
 class SampledSoftmaxLoss(_Loss):
     r"""
     Computes the sampled softmax training loss.

mindspore/nn/metrics/dice.py

@@ -26,7 +26,7 @@ class Dice(Metric):
         The function is shown as follows:
 
         .. math::
-            \text{dice} = \frac{2 * (\text{pred} \bigcap \text{true})}{\text{pred} \bigcup \text{true}}
+            dice = \frac{2 * (pred \bigcap true)}{pred \bigcup true}
 
         Args:
             smooth (float): A term added to the denominator to improve numerical stability. Should be greater than 0.
@@ -58,7 +58,7 @@ class Dice(Metric):
 
     def update(self, *inputs):
         """
-        Updates the internal evaluation result :math:`y_{pred}` and :math:`y`.
+        Updates the internal evaluation result :math:`y_pred` and :math:`y`.
 
         Args:
             inputs: Input `y_pred` and `y`. `y_pred` and `y` are Tensor, list or numpy.ndarray. `y_pred` is the

mindspore/nn/metrics/hausdorff_distance.py

@@ -70,9 +70,9 @@ class HausdorffDistance(Metric):
     Given two feature sets A and B, the Hausdorff distance between two point sets A and B is defined as follows:
 
     .. math::
-        \text{H}(A, B) = \text{max}[\text{h}(A, B), \text{h}(B, A)]
+        H(A, B) = \text{max}[h(A, B), h(B, A)]
-        \text{h}(A, B) = \underset{a \in A}{\text{max}}\{\underset{b \in B}{\text{min}} \rVert a - b \rVert \}
+        h(A, B) = \underset{a \in A}{\text{max}}\{\underset{b \in B}{\text{min}} \rVert a - b \rVert \}
-        \text{h}(A, B) = \underset{b \in B}{\text{max}}\{\underset{a \in A}{\text{min}} \rVert b - a \rVert \}
+        h(A, B) = \underset{b \in B}{\text{max}}\{\underset{a \in A}{\text{min}} \rVert b - a \rVert \}
 
     Args:
         distance_metric (string): The parameter of calculating Hausdorff distance supports three measurement methods,

model_zoo/official/cv/unet/eval.py

@@ -85,6 +85,7 @@ class dice_coeff(nn.Metric):
             raise RuntimeError('Total samples num must not be 0.')
         return self._dice_coeff_sum / float(self._samples_num)
 
+
 def test_net(data_dir,
              ckpt_path,
              cross_valid_ind=1,
@@ -102,6 +103,7 @@ def test_net(data_dir,
     dice_score = model.eval(valid_dataset, dataset_sink_mode=False)
     print("============== Cross valid dice coeff is:", dice_score)
 
+
 def get_args():
     parser = argparse.ArgumentParser(description='Test the UNet on images and target masks',
                                      formatter_class=argparse.ArgumentDefaultsHelpFormatter)

tests/ut/python/nn/test_loss.py

@@ -14,7 +14,8 @@
 # ============================================================================
 """ test loss """
 import numpy as np
-
+import pytest
+import mindspore.common.dtype as mstype
 import mindspore.nn as nn
 from mindspore import Tensor
 from ..ut_filter import non_graph_engine
@@ -88,3 +89,22 @@ def test_cosine_embedding_loss():
     x2 = Tensor(np.array([[0.4, 1.2], [-0.4, -0.9]]).astype(np.float32))
     label = Tensor(np.array([1, -1]).astype(np.int32))
     loss(x1, x2, label)
+
+
+def test_dice_loss():
+    """ test_dice_loss """
+    loss = nn.DiceLoss()
+    y_pred = Tensor(np.array([[0.2, 0.5], [0.3, 0.1], [0.9, 0.6]]), mstype.float32)
+    y = Tensor(np.array([[0, 1], [1, 0], [0, 1]]), mstype.float32)
+    # Pass the test if no error is reported
+    loss(y_pred, y).asnumpy()
+
+
+
+def test_dice_loss_check_shape():
+    """ test_dice_loss """
+    loss = nn.DiceLoss()
+    y_pred = Tensor(np.array([[0.2, 0.5], [0.3, 0.1], [0.9, 0.6]]), mstype.float32)
+    y = Tensor(np.array([[1, 0], [0, 1]]), mstype.float32)
+    with pytest.raises(ValueError):
+        loss(y_pred, y)