add paddle.nn.unfold #32297 (#32298)

* add paddle.nn.unfold
* update Parameters of Unfold

python/paddle/fluid/tests/unittests/test_unfold_op.py

@@ -18,6 +18,9 @@ import math
 import numpy as np
 import unittest
 from op_test import OpTest
+import paddle
+import paddle.fluid as fluid
+from paddle.fluid import core
 
 
 class TestUnfoldOp(OpTest):
@@ -98,5 +101,30 @@ class TestUnfoldOp(OpTest):
         self.check_grad(['X'], 'Y')
 
 
+class TestUnfoldAPI(TestUnfoldOp):
+    """
+    This is for test on paddle.nn.Unfold
+    """
+
+    def setUp(self):
+        self.op_type = 'unfold'
+        self.set_data()
+        self.places = [fluid.CPUPlace()]
+        if core.is_compiled_with_cuda():
+            self.places.append(fluid.CUDAPlace(0))
+
+    def test_dygraph(self):
+        for place in self.places:
+            with fluid.dygraph.guard(place):
+                input = fluid.dygraph.to_variable(self.inputs['X'])
+                m = paddle.nn.Unfold(**self.attrs)
+                m.eval()
+                result = m(input)
+                self.assertTrue(np.allclose(result.numpy(), self.outputs['Y']))
+
+    def test_info(self):
+        str(paddle.nn.Unfold(**self.attrs))
+
+
 if __name__ == '__main__':
     unittest.main()

python/paddle/nn/__init__.py

@@ -83,6 +83,7 @@ from .layer.common import Dropout  #DEFINE_ALIAS
 from .layer.common import Dropout2D  #DEFINE_ALIAS
 from .layer.common import Dropout3D  #DEFINE_ALIAS
 from .layer.common import AlphaDropout  #DEFINE_ALIAS
+from .layer.common import Unfold  #DEFINE_ALIAS
 
 from .layer.pooling import AvgPool1D  #DEFINE_ALIAS
 from .layer.pooling import AvgPool2D  #DEFINE_ALIAS

python/paddle/nn/layer/common.py

@@ -35,6 +35,7 @@ __all__ = [
     'Dropout3D',
     'Bilinear',
     'AlphaDropout',
+    'Unfold',
 ]
 
 
@@ -1380,3 +1381,73 @@ class Embedding(layers.Layer):
         if self._name is not None:
             main_str += ', name={_name}'
         return main_str.format(**self.__dict__)
+
+
+class Unfold(layers.Layer):
+    """
+    This op returns a col buffer of sliding local blocks of input x, also known
+    as im2col for batched 2D image tensors. For each block under the convolution filter,
+    all element will be rearranged as a column. While the convolution filter sliding over
+    the input feature map, a series of such columns will be formed.
+
+    For each input :math:`x` with shape [N, C, H, W], the output shape [N, Cout, Lout]
+    can be calculated as following.
+
+    See ``paddle.nn.functional.unfold`` for more details.
+
+    
+    Parameters:
+        kernel_sizes(int|list):   The size of convolution kernel, should be [k_h, k_w]
+                                  or an integer k treated as [k, k].
+        strides(int|list):        The strides, should be [stride_h, stride_w]
+                                  or an integer stride treated as [sride, stride].
+                                  For default, strides will be [1, 1].
+        paddings(int|list):       The paddings of each dimension, should be
+                                  [padding_top, padding_left, padding_bottom, padding_right]
+                                  or [padding_h, padding_w] or an integer padding.
+                                  If [padding_h, padding_w] was given, it will expanded to
+                                  [padding_h, padding_w, padding_h, padding_w]. If an integer
+                                  padding was given, [padding, padding, padding, padding] will
+                                  be used. For default, paddings will be [0, 0, 0, 0]
+        dilations(int|list):      the dilations of convolution kernel, should be
+                                  [dilation_h, dilation_w], or an integer dilation treated as
+                                  [dilation, dilation]. For default, it will be [1, 1].
+        name(str, optional): The default value is None.
+                             Normally there is no need for user to set this property.
+                             For more information, please refer to :ref:`api_guide_Name`
+
+
+    Examples:
+        .. code-block:: python
+
+            import paddle
+            import paddle.nn as nn
+
+            x = paddle.randn((100,3,224,224))
+            unfold = nn.Unfold(kernel_sizes=[3, 3])
+            result = unfold(x)
+            print(result)
+   """
+
+    def __init__(self,
+                 kernel_sizes,
+                 dilations=1,
+                 paddings=0,
+                 strides=1,
+                 name=None):
+        super(Unfold, self).__init__()
+
+        self.kernel_sizes = kernel_sizes
+        self.dilations = dilations
+        self.paddings = paddings
+        self.strides = strides
+        self.name = name
+
+    def forward(self, input):
+        return F.unfold(input, self.kernel_sizes, self.dilations, self.paddings,
+                        self.strides, self.name)
+
+    def extra_repr(self):
+        name_str = ', name={}'.format(self.name) if self.name else ''
+        return 'kernel_size={}, dilation={}, padding={}, stride={}{}'.\
+                format(self.kernel_sizes, self.dilations, self.paddings, self.strides, name_str)