Ctrl+K

depthwise_conv2d#

ivy.depthwise_conv2d(x, filters, strides, padding, /, *, data_format='NHWC', dilations=1, out=None)[source]#

Compute a 2-D depthwise convolution given 4-D input x and filters arrays.

Parameters:

  • x (Union[Array, NativeArray]) – Input image [batch_size,h,w,d_in] or [batch_size,d_in,h,w].

  • filters (Union[Array, NativeArray]) – Convolution filters [fh,fw,d_in]. (d_in must be the same as d from x)

  • strides (Union[int, Tuple[int, int]]) – The stride of the sliding window for each dimension of input.

  • padding (Union[str, Sequence[Tuple[int, int]]]) – either the string ‘SAME’ (padding with zeros evenly), the string ‘VALID’ (no padding), or a sequence of n (low, high) integer pairs that give the padding to apply before and after each spatial dimension.

  • data_format (str, default: 'NHWC') – The ordering of the dimensions in the input, one of “NHWC” or “NCHW”. “NHWC” corresponds to inputs with shape (batch_size, height, width, channels), while “NCHW” corresponds to input with shape (batch_size, channels, height, width).

  • dilations (Union[int, Tuple[int, int]], default: 1) – The dilation factor for each dimension of input. (Default value = 1)

  • out (Optional[Array], default: None) – optional output array, for writing the result to. It must have a shape that the inputs broadcast to.

Return type:

Array

Returns:

  • ret – The result of the convolution operation.

  • Both the description and the type hints above assumes an array input for simplicity,

  • but this function is nestable, and therefore also accepts ivy.Container

  • instances in place of any of the arguments.

Examples

With ivy.Array input:

>>> x = ivy.random_normal(mean=0, std=1, shape=[1, 28, 28, 3])
>>> filters = ivy.random_normal(mean=0, std=1, shape=[3, 3, 3])
>>> y = ivy.depthwise_conv2d(x, filters, (1, 1), 'VALID')
>>> print(y.shape)
ivy.Shape(1, 26, 26, 3)

>>> x = ivy.random_normal(mean=0, std=1, shape=[1, 32, 32, 3])
>>> y = ivy.zeros((1, 16, 16, 3))
>>> filters = ivy.random_normal(mean=0, std=1, shape=[5, 5, 3])
>>> ivy.depthwise_conv2d(x, filters, [2, 2], 'SAME', out=y)
>>> print(y.shape)
ivy.Shape(1, 16, 16, 3)

>>> x = ivy.random_normal(mean=0, std=1, shape=[1, 64, 64, 32])
>>> y = ivy.zeros((1, 61, 61, 32))
>>> filters = ivy.random_normal(mean=0, std=1, shape=[4, 4, 32])
>>> ivy.depthwise_conv2d(x, filters, [1, 1], 'VALID', out=y)
>>> print(x.shape)
ivy.Shape(1, 64, 64, 32)

With ivy.NativeArray input:

>>> x = ivy.native_array(ivy.random_normal(mean=0, std=1, shape=[1, 7, 7, 64]))
>>> filters = ivy.native_array(ivy.random_normal(mean=0, std=1, shape=[3, 3, 64]))
>>> y = ivy.depthwise_conv2d(x, filters, [1, 1], 'SAME')
>>> print(y.shape)
ivy.Shape(1, 7, 7, 64)

With a mix of ivy.Array and ivy.Container inputs:

>>> x = ivy.eye(6, 6).reshape((1, 6, 6, 1)) #NHWC
>>> a = ivy.array([[1., 1., 1.], [1., -8., 1.], [1., 1., 1.]]).expand_dims(axis=-1)
>>> b = ivy.array([[1., 1., 1.],
...                [1., 1., 1.],
...                [1., 1., 1.]]).expand_dims(axis=-1) / 9.0
>>> filters = ivy.Container(a = a, b = b)
>>> y = ivy.depthwise_conv2d(x, filters, 1, 'VALID', dilations=2)
>>> print(y)
{
    a: ivy.array([[[[-6.],
                    [0.]],
                   [[0.],
                    [-6.]]]]),
    b: ivy.array([[[[0.33333334],
                    [0.]],
                   [[0.],
                    [0.33333334]]]])
}

With a mix of ivy.Array, code:ivy.NativeArray and ivy.Container inputs:

>>> x = ivy.eye(6, 6).reshape((1, 6, 6, 1)) #NHWC
>>> y = ivy.native_array(ivy.eye(6, 6).reshape((1, 6, 6, 1)))
>>> inp = ivy.Container(x = x, y = y)
>>> filter = ivy.array([[1., 1., 1.],
...                     [1., -8., 1.],
...                     [1., 1., 1.]]).expand_dims(axis=-1)
>>> y = ivy.depthwise_conv2d(inp, filter, 1, 'VALID', dilations=2)
>>> print(y)
{
    x: ivy.array([[[[-6.],
                    [0.]],
                   [[0.],
                    [-6.]]]]),
    y: ivy.array([[[[-6.],
                    [0.]],
                   [[0.],
                    [-6.]]]])
}
Array.depthwise_conv2d(self, filters, strides, padding, /, *, data_format='NHWC', dilations=1, out=None)[source]#

ivy.Array instance method variant of ivy.depthwise_conv2d. This method simply wraps the function, and so the docstring for ivy.depthwise_conv2d also applies to this method with minimal changes.

Parameters:

  • self (Array) – Input image [batch_size,h,w,d].

  • filters (Union[Array, NativeArray]) – Convolution filters [fh,fw,d_in]. (d_in must be the same as d from self)

  • strides (Union[int, Tuple[int], Tuple[int, int]]) – The stride of the sliding window for each dimension of input.

  • padding (Union[str, List[int]]) – “SAME” or “VALID” indicating the algorithm, or list indicating the per-dimension paddings.

  • data_format (str, default: 'NHWC') – “NHWC” or “NCHW”. Defaults to “NHWC”.

  • dilations (Union[int, Tuple[int], Tuple[int, int]], default: 1) – The dilation factor for each dimension of input. (Default value = 1)

  • out (Optional[Array], default: None) – optional output array, for writing the result to. It must have a shape that the inputs broadcast to.

Return type:

Array

Returns:

ret – The result of the convolution operation.

Examples

>>> x = ivy.randint(0, 255, shape=(1, 128, 128, 3)).astype(ivy.float32) / 255.0
>>> filters = ivy.random_normal(mean=0, std=1, shape=[3, 3, 3])
>>> y = x.depthwise_conv2d(filters, 2, 'SAME')
>>> print(y.shape)
(1, 64, 64, 3)
Container.depthwise_conv2d(self, filters, strides, padding, /, *, data_format='NHWC', dilations=1, key_chains=None, to_apply=True, prune_unapplied=False, map_sequences=False, out=None)[source]#

ivy.Container instance method variant of ivy.depthwise_conv2d. This method simply wraps the function, and so the docstring for ivy.depthwise_conv2d also applies to this method with minimal changes.

Parameters:

  • self (Container) – Input image [batch_size,h,w,d].

  • filters (Union[Array, NativeArray, Container]) – Convolution filters [fh,fw,d_in]. (d_in must be the same as d from self)

  • strides (Union[int, Tuple[int], Tuple[int, int], Container]) – The stride of the sliding window for each dimension of input.

  • padding (Union[str, List[int], Container]) – “SAME” or “VALID” indicating the algorithm, or list indicating the per-dimension paddings.

  • data_format (Union[str, Container], default: 'NHWC') – “NHWC” or “NCHW”. Defaults to “NHWC”.

  • dilations (Union[int, Tuple[int], Tuple[int, int], Container], default: 1) – The dilation factor for each dimension of input. (Default value = 1)

  • out (Optional[Container], default: None) – optional output container, for writing the result to. It must have a shape that the inputs broadcast to.

Return type:

Container

Returns:

ret – The result of the convolution operation.

Examples

>>> a = ivy.randint(0, 255, shape=(1, 128, 128, 3)).astype(ivy.float32) / 255.0
>>> b = ivy.randint(0, 255, shape=(1, 128, 128, 3)).astype(ivy.float32) / 255.0
>>> inp = ivy.Container(a=a, b=b)
>>> filters = ivy.random_normal(mean=0, std=1, shape=[3, 3, 3])
>>> y = inp.depthwise_conv2d(filters, 2, 'SAME')
>>> print(y.shape)
[1, 64, 64, 3]
On this page