!11813 modify dice and diceloss

From: @lijiaqi0612
Reviewed-by: @kingxian,@zh_qh
Signed-off-by: @kingxian

mindspore/nn/loss/loss.py

@@ -310,11 +310,10 @@ class DiceLoss(_Loss):
     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).
+        - **y_pred** (Tensor) - Tensor of shape (N, ...).
+        - **y** (Tensor) - Tensor of shape (N, ...).
 
     Outputs:
         Tensor, a tensor of shape with the per-example sampled Dice losses.
@@ -323,32 +322,30 @@ class DiceLoss(_Loss):
         ``Ascend``
 
     Examples:
-        >>> loss = nn.Diceloss(smooth=1e-5, threshold=0.5)
+        >>> loss = nn.DiceLoss(smooth=1e-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]
+        [0.7953220862819745]
+
+    Raises:
+        ValueError: If the dimensions are different.
+        TypeError: If the type of inputs are not Tensor.
     """
-    def __init__(self, smooth=1e-5, threshold=0.5):
+    def __init__(self, smooth=1e-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(self.mul(logits.view(-1), label.view(-1)))
+        unionset = self.reduce_sum(self.mul(logits.view(-1), logits.view(-1))) + \
+                   self.reduce_sum(self.mul(label.view(-1), label.view(-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]
+        single_dice_coeff = (2 * intersection) / (unionset + self.smooth)
+        dice_loss = 1 - single_dice_coeff / label.shape[0]
 
         return dice_loss
 

mindspore/nn/metrics/dice.py

@@ -26,35 +26,35 @@ class Dice(Metric):
     The function is shown as follows:
 
     .. math::
-            dice = \frac{2 * (pred \bigcap true)}{pred \bigcup 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.
                         Default: 1e-5.
-        threshold (float): A threshold, which is used to compare with the input tensor. Default: 0.5.
 
     Examples:
         >>> x = Tensor(np.array([[0.2, 0.5], [0.3, 0.1], [0.9, 0.6]]))
         >>> y = Tensor(np.array([[0, 1], [1, 0], [0, 1]]))
-        >>> metric = Dice(smooth=1e-5, threshold=0.5)
+        >>> metric = Dice(smooth=1e-5)
         >>> metric.clear()
         >>> metric.update(x, y)
         >>> dice = metric.eval()
-        0.22222926
+        >>> print(dice)
+        0.20467791371802546
     """
 
-    def __init__(self, smooth=1e-5, threshold=0.5):
+    def __init__(self, smooth=1e-5):
         super(Dice, self).__init__()
 
         self.smooth = validator.check_positive_float(smooth, "smooth")
-        self.threshold = validator.check_value_type("threshold", threshold, [float])
+        self._dice_coeff_sum = 0
+        self._samples_num = 0
         self.clear()
 
     def clear(self):
         """Clears the internal evaluation result."""
-        self._dim = 0
-        self.intersection = 0
-        self.unionset = 0
+        self._dice_coeff_sum = 0
+        self._samples_num = 0
 
     def update(self, *inputs):
         """
@@ -62,7 +62,7 @@ class Dice(Metric):
 
         Args:
             inputs: Input `y_pred` and `y`. `y_pred` and `y` are Tensor, list or numpy.ndarray. `y_pred` is the
-                    predicted value, `y` is the true value. The shape of `y_pred` and `y` are both :math:`(N, C)`.
+                    predicted value, `y` is the true value. The shape of `y_pred` and `y` are both :math:`(N, ...)`.
 
         Raises:
             ValueError: If the number of the inputs is not 2.
@@ -72,17 +72,17 @@ class Dice(Metric):
 
         y_pred = self._convert_data(inputs[0])
         y = self._convert_data(inputs[1])
+        self._samples_num += y.shape[0]
 
         if y_pred.shape != y.shape:
             raise RuntimeError('y_pred and y should have same the dimension, but the shape of y_pred is{}, '
                                'the shape of y is {}.'.format(y_pred.shape, y.shape))
 
-        y_pred = (y_pred > self.threshold).astype(int)
-        self._dim = y.shape
-        pred_flat = np.reshape(y_pred, (self._dim[0], -1))
-        true_flat = np.reshape(y, (self._dim[0], -1))
-        self.intersection = np.sum((pred_flat * true_flat), axis=1)
-        self.unionset = np.sum(pred_flat, axis=1) + np.sum(true_flat, axis=1)
+        intersection = np.dot(y_pred.flatten(), y.flatten())
+        unionset = np.dot(y_pred.flatten(), y_pred.flatten()) + np.dot(y.flatten(), y.flatten())
+
+        single_dice_coeff = 2 * float(intersection) / float(unionset + self.smooth)
+        self._dice_coeff_sum += single_dice_coeff
 
     def eval(self):
         r"""
@@ -92,11 +92,9 @@ class Dice(Metric):
             Float, the computed result.
 
         Raises:
-            RuntimeError: If the sample size is 0.
+            RuntimeError: If the total samples num is 0.
         """
-        if self._dim[0] == 0:
-            raise RuntimeError('Dice can not be calculated, because the number of samples is 0.')
+        if self._samples_num == 0:
+            raise RuntimeError('Total samples num must not be 0.')
 
-        dice = (2 * self.intersection + self.smooth) / (self.unionset + self.smooth)
-
-        return np.sum(dice) / self._dim[0]
+        return self._dice_coeff_sum / float(self._samples_num)

tests/ut/python/metrics/test_dice.py

@@ -29,12 +29,12 @@ def test_classification_dice():
     metric.update(x, y)
     dice = metric.eval()
 
-    assert math.isclose(dice, 0.22222926, abs_tol=0.001)
+    assert math.isclose(dice, 0.20467791371802546, abs_tol=0.001)
 
 
 def test_dice_update1():
     x = Tensor(np.array([[0.2, 0.5, 0.7], [0.3, 0.1, 0.2], [0.9, 0.6, 0.5]]))
-    metric = Dice(1e-5, 0.5)
+    metric = Dice(1e-5)
     metric.clear()
 
     with pytest.raises(ValueError):
@@ -42,8 +42,8 @@ def test_dice_update1():
 
 
 def test_dice_runtime():
-    metric = Dice(1e-5, 0.8)
+    metric = Dice(1e-5)
     metric.clear()
 
-    with pytest.raises(TypeError):
+    with pytest.raises(RuntimeError):
         metric.eval()

tests/ut/python/nn/test_loss.py

@@ -97,8 +97,7 @@ def test_dice_loss():
     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()
-
+    loss(y_pred, y)
 
 
 def test_dice_loss_check_shape():