diff --git a/doc/fluid/api/nn.rst b/doc/fluid/api/nn.rst index f88d8a2918faff378fd1dde0e5afb7f725c16a3a..b8575ed2df757532889dd0bb818106c971e9d564 100644 --- a/doc/fluid/api/nn.rst +++ b/doc/fluid/api/nn.rst @@ -154,4 +154,6 @@ paddle.nn nn/functional/loss/margin_ranking_loss.rst nn/functional/activation/sigmoid.rst nn/layer/loss/MarginRankingLoss.rst - nn/layer/activation/Sigmoid.rst + nn/AdaptiveAvgPool2d.rst + nn/AdaptiveAvgPool3d.rst + nn/layer/activation/Sigmoid.rst \ No newline at end of file diff --git a/doc/fluid/api/nn/AdaptiveAvgPool2d.rst b/doc/fluid/api/nn/AdaptiveAvgPool2d.rst new file mode 100644 index 0000000000000000000000000000000000000000..26518d01d0e432d262e4434ced097fbf12e4a117 --- /dev/null +++ b/doc/fluid/api/nn/AdaptiveAvgPool2d.rst @@ -0,0 +1,10 @@ +.. _api_nn_pooling_AdaptiveAvgPool2d: + +AdaptiveAvgPool2d +----------------- + +.. autoclass:: paddle.nn.AdaptiveAvgPool2d + :members: + :inherited-members: + :noindex: + diff --git a/doc/fluid/api/nn/AdaptiveAvgPool3d.rst b/doc/fluid/api/nn/AdaptiveAvgPool3d.rst new file mode 100644 index 0000000000000000000000000000000000000000..70f9e87f2b59bfbf7f754e4a615c49fa85fb6b4f --- /dev/null +++ b/doc/fluid/api/nn/AdaptiveAvgPool3d.rst @@ -0,0 +1,10 @@ +.. _api_nn_pooling_AdaptiveAvgPool3d: + +AdaptiveAvgPool3d +----------------- + +.. autoclass:: paddle.nn.AdaptiveAvgPool3d + :members: + :inherited-members: + :noindex: + diff --git a/doc/fluid/api/nn/functional.rst b/doc/fluid/api/nn/functional.rst index 598b76a479602a53a4d7073bc31c65ba3eddbe53..25d0a9743a51d6813fba6bf15a4e1e4118e8a469 100644 --- a/doc/fluid/api/nn/functional.rst +++ b/doc/fluid/api/nn/functional.rst @@ -8,3 +8,5 @@ functional functional/l1_loss.rst functional/nll_loss.rst functional/mse_loss.rst + functional/adaptive_avg_pool2d.rst + functional/adaptive_avg_pool3d.rst diff --git a/doc/fluid/api/nn/functional/adaptive_avg_pool2d.rst b/doc/fluid/api/nn/functional/adaptive_avg_pool2d.rst new file mode 100644 index 0000000000000000000000000000000000000000..d0eff20c18b8ea6d58cfe0405c16aee6f721d30f --- /dev/null +++ b/doc/fluid/api/nn/functional/adaptive_avg_pool2d.rst @@ -0,0 +1,8 @@ +.. _api_nn_functional_adaptive_avg_pool2d: + +adaptive_avg_pool2d +-------------------- + +.. autofunction:: paddle.nn.functional.adaptive_avg_pool2d + :noindex: + diff --git a/doc/fluid/api/nn/functional/adaptive_avg_pool3d.rst b/doc/fluid/api/nn/functional/adaptive_avg_pool3d.rst new file mode 100644 index 0000000000000000000000000000000000000000..4765c3e11179f685033023fa8e7cd50846367d55 --- /dev/null +++ b/doc/fluid/api/nn/functional/adaptive_avg_pool3d.rst @@ -0,0 +1,8 @@ +.. _api_nn_functional_adaptive_avg_pool3d: + +adaptive_avg_pool3d +-------------------- + +.. autofunction:: paddle.nn.functional.adaptive_avg_pool3d + :noindex: + diff --git a/doc/fluid/api_cn/nn_cn.rst b/doc/fluid/api_cn/nn_cn.rst index fccb477acfe1467ff10170780012419298ac7e3f..31926805f9303f1669fe1cd9e67a1b6a0a4baae1 100644 --- a/doc/fluid/api_cn/nn_cn.rst +++ b/doc/fluid/api_cn/nn_cn.rst @@ -166,4 +166,6 @@ paddle.nn nn_cn/yolo_box_cn.rst nn_cn/loss_cn/MarginRankingLoss_cn.rst nn_cn/functional_cn/margin_ranking_loss_cn.rst + nn_cn/AdaptiveAvgPool2d_cn.rst + nn_cn/AdaptiveAvgPool3d_cn.rst diff --git a/doc/fluid/api_cn/nn_cn/AdaptiveAvgPool2d_cn.rst b/doc/fluid/api_cn/nn_cn/AdaptiveAvgPool2d_cn.rst new file mode 100755 index 0000000000000000000000000000000000000000..6d6eaa5044f9b2781c46f449240e8bf158c3fb50 --- /dev/null +++ b/doc/fluid/api_cn/nn_cn/AdaptiveAvgPool2d_cn.rst @@ -0,0 +1,72 @@ +AdaptiveAvgPool2d +------------------------------- + +.. py:function:: paddle.nn.AdaptiveAvgPool2d(output_size, data_format="NCHW", name=None) + +该算子根据输入 `x` , `output_size` 等参数对一个输入Tensor计算2D的自适应平均池化。输入和输出都是4-D Tensor， +默认是以 `NCHW` 格式表示的，其中 `N` 是 batch size, `C` 是通道数, `H` 是输入特征的高度, `H` 是输入特征的宽度。 + +计算公式如下: + +.. math:: + + hstart &= floor(i * H_{in} / H_{out}) + + hend &= ceil((i + 1) * H_{in} / H_{out}) + + wstart &= floor(j * W_{in} / W_{out}) + + wend &= ceil((j + 1) * W_{in} / W_{out}) + + Output(i ,j) &= \frac{sum(Input[hstart:hend, wstart:wend])}{(hend - hstart) * (wend - wstart)} + + +参数 +::::::::: + - **output_size** (int|list|turple): 算子输出特征图的尺寸，如果其是list或turple类型的数值，必须包含两个元素，H和W。H和W既可以是int类型值也可以是None，None表示与输入特征尺寸相同。 + - **data_format** (str): 输入和输出的数据格式，可以是"NCHW"和"NHWC"。N是批尺寸，C是通道数，H是特征高度，W是特征宽度。默认值："NCHW"。 + - **name** (str，可选): 操作的名称(可选，默认值为None）。更多信息请参见 :ref:`api_guide_Name`。 + +形状 +::::::::: + - **x** (Tensor): 默认形状为（批大小，通道数，高度，宽度），即NCHW格式的4-D Tensor。 其数据类型为float16, float32, float64, int32或int64。 + - **output** (Tensor): 默认形状为（批大小，通道数，输出特征高度，输出特征宽度），即NCHW格式的4-D Tensor。 其数据类型与输入相同。 + + +返回 +::::::::: +计算AdaptiveAvgPool2d的可调用对象 + +抛出异常 +::::::::: + - ``ValueError`` - 如果 ``data_format`` 既不是"NCHW"也不是"NHWC"。 + +代码示例 +::::::::: + +.. code-block:: python + + # adaptive avg pool2d + # suppose input data in shape of [N, C, H, W], `output_size` is [m, n], + # output shape is [N, C, m, n], adaptive pool divide H and W dimensions + # of input data into m * n grids averagely and performs poolings in each + # grid to get output. + # adaptive avg pool performs calculations as follow: + # + # for i in range(m): + # for j in range(n): + # hstart = floor(i * H / m) + # hend = ceil((i + 1) * H / m) + # wstart = floor(i * W / n) + # wend = ceil((i + 1) * W / n) + # output[:, :, i, j] = avg(input[:, :, hstart: hend, wstart: wend]) + # + import paddle + import numpy as np + paddle.disable_static() + input_data = np.random.rand(2, 3, 32, 32) + x = paddle.to_tensor(input_data) + # x.shape is [2, 3, 32, 32] + adaptive_avg_pool = paddle.nn.AdaptiveAvgPool2d(output_size=3) + pool_out = adaptive_avg_pool(x = x) + # pool_out.shape is [2, 3, 3, 3] \ No newline at end of file diff --git a/doc/fluid/api_cn/nn_cn/AdaptiveAvgPool3d_cn.rst b/doc/fluid/api_cn/nn_cn/AdaptiveAvgPool3d_cn.rst new file mode 100755 index 0000000000000000000000000000000000000000..4315f960cdf53e6f6f25c1b06d600d84d3b03dd2 --- /dev/null +++ b/doc/fluid/api_cn/nn_cn/AdaptiveAvgPool3d_cn.rst @@ -0,0 +1,78 @@ +AdaptiveAvgPool3d +------------------------------- + +.. py:function:: paddle.nn.AdaptiveAvgPool3d(output_size, data_format="NCDHW", name=None) + +该算子根据输入 `x` , `output_size` 等参数对一个输入Tensor计算3D的自适应平均池化。输入和输出都是5-D Tensor， +默认是以 `NCDHW` 格式表示的，其中 `N` 是 batch size, `C` 是通道数, `D` 是特征图长度, `H` 是输入特征的高度, `H` 是输入特征的宽度。 + +计算公式如下: + +.. math:: + + dstart &= floor(i * D_{in} / D_{out}) + + dend &= ceil((i + 1) * D_{in} / D_{out}) + + hstart &= floor(j * H_{in} / H_{out}) + + hend &= ceil((j + 1) * H_{in} / H_{out}) + + wstart &= floor(k * W_{in} / W_{out}) + + wend &= ceil((k + 1) * W_{in} / W_{out}) + + Output(i ,j, k) &= \frac{sum(Input[dstart:dend, hstart:hend, wstart:wend])}{(dend - dstart) * (hend - hstart) * (wend - wstart)} + +参数 +::::::::: + - **output_size** (int|list|turple): 算子输出特征图的尺寸，如果其是list或turple类型的数值，必须包含三个元素，D，H和W。D，H和W既可以是int类型值也可以是None，None表示与输入特征尺寸相同。 + - **data_format** (str): 输入和输出的数据格式，可以是"NCDHW"和"NDHWC"。N是批尺寸，C是通道数，D是特征长度，H是特征高度，W是特征宽度。默认值："NCDHW"。 + - **name** (str，可选): 操作的名称(可选，默认值为None）。更多信息请参见 :ref:`api_guide_Name`。 + +形状 +::::::::: + - **x** (Tensor): 默认形状为（批大小，通道数，长度，高度，宽度），即NCDHW格式的5-D Tensor。 其数据类型为float16, float32, float64, int32或int64. + - **output** (Tensor): 默认形状为（批大小，通道数，输出特征长度，输出特征高度，输出特征宽度），即NCDHW格式的5-D Tensor。 其数据类型与输入相同。 + + +返回 +::::::::: +计算AdaptiveAvgPool3d的可调用对象 + +抛出异常 +::::::::: + - ``ValueError`` - 如果 ``data_format`` 既不是"NCDHW"也不是"NDHWC"。 + +代码示例 +::::::::: + +.. code-block:: python + + # adaptive avg pool3d + # suppose input data in shape of [N, C, D, H, W], `output_size` is [l, m, n], + # output shape is [N, C, l, m, n], adaptive pool divide D, H and W dimensions + # of input data into l * m * n grids averagely and performs poolings in each + # grid to get output. + # adaptive avg pool performs calculations as follow: + # + # for i in range(l): + # for j in range(m): + # for k in range(n): + # dstart = floor(i * D / l) + # dend = ceil((i + 1) * D / l) + # hstart = floor(j * H / m) + # hend = ceil((j + 1) * H / m) + # wstart = floor(k * W / n) + # wend = ceil((k + 1) * W / n) + # output[:, :, i, j, k] = + # avg(input[:, :, dstart:dend, hstart: hend, wstart: wend]) + import paddle + import numpy as np + paddle.disable_static() + input_data = np.random.rand(2, 3, 8, 32, 32) + x = paddle.to_tensor(input_data) + # x.shape is [2, 3, 8, 32, 32] + adaptive_avg_pool = paddle.nn.AdaptiveAvgPool3d(output_size=3) + pool_out = adaptive_avg_pool(x = x) + # pool_out = [2, 3, 3, 3, 3] diff --git a/doc/fluid/api_cn/nn_cn/functional_cn.rst b/doc/fluid/api_cn/nn_cn/functional_cn.rst index d3c9b813bbe0f2858e4b3245b6e3652b04bd43be..bc7af04cb288b94489324930c4a4ae56a4f50968 100644 --- a/doc/fluid/api_cn/nn_cn/functional_cn.rst +++ b/doc/fluid/api_cn/nn_cn/functional_cn.rst @@ -14,4 +14,6 @@ functional functional_cn/margin_ranking_loss_cn.rst functional_cn/sigmoid_cn.rst functional_cn/mse_loss_cn.rst - functional_cn/sigmoid_cn.rst + functional_cn/adaptive_avg_pool2d_cn.rst + functional_cn/adaptive_avg_pool3d_cn.rst + functional_cn/sigmoid_cn.rst \ No newline at end of file diff --git a/doc/fluid/api_cn/nn_cn/functional_cn/adaptive_avg_pool2d_cn.rst b/doc/fluid/api_cn/nn_cn/functional_cn/adaptive_avg_pool2d_cn.rst new file mode 100755 index 0000000000000000000000000000000000000000..cd5e3a087a4e187c9015c2a19a96112295012c15 --- /dev/null +++ b/doc/fluid/api_cn/nn_cn/functional_cn/adaptive_avg_pool2d_cn.rst @@ -0,0 +1,68 @@ +adaptive_avg_pool2d +------------------------------- + +.. py:function:: paddle.nn.functional.adaptive_avg_pool2d(x, output_size, data_format='NCHW', name=None) + +该算子根据输入 `x` , `output_size` 等参数对一个输入Tensor计算2D的自适应平均池化。输入和输出都是4-D Tensor， +默认是以 `NCHW` 格式表示的，其中 `N` 是 batch size, `C` 是通道数, `H` 是输入特征的高度, `H` 是输入特征的宽度。 + +计算公式如下: + +.. math:: + + hstart &= floor(i * H_{in} / H_{out}) + + hend &= ceil((i + 1) * H_{in} / H_{out}) + + wstart &= floor(j * W_{in} / W_{out}) + + wend &= ceil((j + 1) * W_{in} / W_{out}) + + Output(i ,j) &= \frac{sum(Input[hstart:hend, wstart:wend])}{(hend - hstart) * (wend - wstart)} + + +参数 +::::::::: + - **x** (Tensor): 默认形状为（批大小，通道数，高度，宽度），即NCHW格式的4-D Tensor。 其数据类型为float16, float32, float64, int32或int64. + - **output_size** (int|list|turple): 算子输出特征图的尺寸，如果其是list或turple类型的数值，必须包含两个元素，H和W。H和W既可以是int类型值也可以是None，None表示与输入特征尺寸相同。 + - **data_format** (str): 输入和输出的数据格式，可以是"NCHW"和"NHWC"。N是批尺寸，C是通道数，H是特征高度，W是特征宽度。默认值："NCHW"。 + - **name** (str，可选): 操作的名称(可选，默认值为None）。更多信息请参见 :ref:`api_guide_Name`。 + +返回 +::::::::: +``Tensor``, 默认形状为（批大小，通道数，输出特征高度，输出特征宽度），即NCHW格式的4-D Tensor，其数据类型与输入相同。 + +抛出异常 +::::::::: + - ``ValueError`` - 如果 ``data_format`` 既不是"NCHW"也不是"NHWC"。 + +代码示例 +::::::::: + +.. code-block:: python + + # adaptive avg pool2d + # suppose input data in shape of [N, C, H, W], `output_size` is [m, n], + # output shape is [N, C, m, n], adaptive pool divide H and W dimensions + # of input data into m * n grids averagely and performs poolings in each + # grid to get output. + # adaptive avg pool performs calculations as follow: + # + # for i in range(m): + # for j in range(n): + # hstart = floor(i * H / m) + # hend = ceil((i + 1) * H / m) + # wstart = floor(i * W / n) + # wend = ceil((i + 1) * W / n) + # output[:, :, i, j] = avg(input[:, :, hstart: hend, wstart: wend]) + # + import paddle + import numpy as np + paddle.disable_static() + input_data = np.random.rand(2, 3, 32, 32) + x = paddle.to_tensor(input_data) + # x.shape is [2, 3, 32, 32] + pool_out = paddle.nn.functional.adaptive_avg_pool2d( + x = x, + output_size=[3, 3]) + # pool_out.shape is [2, 3, 3, 3] \ No newline at end of file diff --git a/doc/fluid/api_cn/nn_cn/functional_cn/adaptive_avg_pool3d_cn.rst b/doc/fluid/api_cn/nn_cn/functional_cn/adaptive_avg_pool3d_cn.rst new file mode 100755 index 0000000000000000000000000000000000000000..756f9d01ef220e7c42253d2a53572dba5f619f43 --- /dev/null +++ b/doc/fluid/api_cn/nn_cn/functional_cn/adaptive_avg_pool3d_cn.rst @@ -0,0 +1,76 @@ +adaptive_avg_pool3d +------------------------------- + +.. py:function:: paddle.nn.functional.adaptive_avg_pool3d(x, output_size, data_format='NCDHW', name=None) + +该算子根据输入 `x` , `output_size` 等参数对一个输入Tensor计算3D的自适应平均池化。输入和输出都是5-D Tensor， +默认是以 `NCDHW` 格式表示的，其中 `N` 是 batch size, `C` 是通道数, `D` 是特征图长度, `H` 是输入特征的高度, `H` 是输入特征的宽度。 + +计算公式如下: + +.. math:: + + dstart &= floor(i * D_{in} / D_{out}) + + dend &= ceil((i + 1) * D_{in} / D_{out}) + + hstart &= floor(j * H_{in} / H_{out}) + + hend &= ceil((j + 1) * H_{in} / H_{out}) + + wstart &= floor(k * W_{in} / W_{out}) + + wend &= ceil((k + 1) * W_{in} / W_{out}) + + Output(i ,j, k) &= \frac{sum(Input[dstart:dend, hstart:hend, wstart:wend])}{(dend - dstart) * (hend - hstart) * (wend - wstart)} + + + +参数 +::::::::: + - **x** (Tensor): 默认形状为（批大小，通道数，长度，高度，宽度），即NCDHW格式的5-D Tensor。 其数据类型为float16, float32, float64, int32或int64. + - **output_size** (int|list|turple): 算子输出特征图的尺寸，如果其是list或turple类型的数值，必须包含三个元素，D，H和W。D，H和W既可以是int类型值也可以是None，None表示与输入特征尺寸相同。 + - **data_format** (str): 输入和输出的数据格式，可以是"NCDHW"和"NDHWC"。N是批尺寸，C是通道数，D是特征长度，H是特征高度，W是特征宽度。默认值："NCDHW"。 + - **name** (str，可选): 操作的名称(可选，默认值为None）。更多信息请参见 :ref:`api_guide_Name`。 + +返回 +::::::::: +``Tensor``, 默认形状为（批大小，通道数，输出特征长度，输出特征高度，输出特征宽度），即NCDHW格式的5-D Tensor，其数据类型与输入相同。 + +抛出异常 +::::::::: + - ``ValueError`` - 如果 ``data_format`` 既不是"NCDHW"也不是"NDHWC"。 + +代码示例 +::::::::: + +.. code-block:: python + + # adaptive avg pool3d + # suppose input data in shape of [N, C, D, H, W], `output_size` is [l, m, n], + # output shape is [N, C, l, m, n], adaptive pool divide D, H and W dimensions + # of input data into l * m * n grids averagely and performs poolings in each + # grid to get output. + # adaptive avg pool performs calculations as follow: + # + # for i in range(l): + # for j in range(m): + # for k in range(n): + # dstart = floor(i * D / l) + # dend = ceil((i + 1) * D / l) + # hstart = floor(j * H / m) + # hend = ceil((j + 1) * H / m) + # wstart = floor(k * W / n) + # wend = ceil((k + 1) * W / n) + # output[:, :, i, j, k] = + # avg(input[:, :, dstart:dend, hstart: hend, wstart: wend]) + import paddle + import numpy as np + paddle.disable_static() + input_data = np.random.rand(2, 3, 8, 32, 32) + x = paddle.to_tensor(input_data) + # x.shape is [2, 3, 8, 32, 32] + pool_out = paddle.nn.functional.adaptive_avg_pool3d( + x = x, + output_size=[3, 3, 3]) + # pool_out.shape is [2, 3, 3, 3, 3] \ No newline at end of file