Fix pool2d issues. Add test case. (#22)

luo-cheng2021 · May 13, 2021 · f01ceb8 · f01ceb8
1 parent 8e8b374
commit f01ceb8
Show file tree

Hide file tree

Showing 4 changed files with 60 additions and 41 deletions.
diff --git a/ngraph/frontend/paddlepaddle/src/op/pool2d.cpp b/ngraph/frontend/paddlepaddle/src/op/pool2d.cpp
@@ -1,17 +1,7 @@
 //*****************************************************************************
-// Copyright 2017-2021 Intel Corporation
+// Copyright (C) 2018-2021 Intel Corporation
+// SPDX-License-Identifier: Apache-2.0
 //
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
 //*****************************************************************************
 
 #include <ngraph/opsets/opset6.hpp>
@@ -47,30 +37,16 @@ static void get_paddings(const NodeContext& node, ngraph::Shape& pad_begin, ngra
     auto paddings = node.get_attribute<std::vector<int32_t>>("paddings");
     auto data_format = node.get_attribute<std::string>("data_format");
 
+    // TODO: need to support NHWC input
     switch (paddings.size())
     {
-    case 1:
-        pad_begin = Shape(2, paddings[0]);
-        pad_end = pad_begin;
-        break;
     case 2:
         pad_begin = Shape{static_cast<uint64_t>(paddings[0]), static_cast<uint64_t>(paddings[1])};
         pad_end = pad_begin;
         break;
     case 4:
         pad_begin = Shape{static_cast<uint64_t>(paddings[0]), static_cast<uint64_t>(paddings[2])};
-        pad_end = Shape(static_cast<uint64_t>(paddings[1]), static_cast<uint64_t>(paddings[3]));
-        break;
-    case 8:
-        if (data_format == "NCHW") {
-            pad_begin = Shape{static_cast<uint64_t>(paddings[4]), static_cast<uint64_t>(paddings[6])};
-            pad_end = Shape(static_cast<uint64_t>(paddings[5]), static_cast<uint64_t>(paddings[7]));
-        } else if (data_format == "NHWC") {
-            pad_begin = Shape{static_cast<uint64_t>(paddings[2]), static_cast<uint64_t>(paddings[4])};
-            pad_end = Shape(static_cast<uint64_t>(paddings[3]), static_cast<uint64_t>(paddings[5]));
-        } else {
-            throw std::runtime_error("Unsupported pooling data_format " + data_format);
-        }
+        pad_end = Shape{static_cast<uint64_t>(paddings[1]), static_cast<uint64_t>(paddings[3])};
         break;
     default:
         throw std::runtime_error("Unsupported pooling paddings " + paddings.size());

diff --git a/ngraph/frontend/paddlepaddle/src/op/pool2d.hpp b/ngraph/frontend/paddlepaddle/src/op/pool2d.hpp
@@ -1,17 +1,7 @@
 //*****************************************************************************
-// Copyright 2017-2021 Intel Corporation
+// Copyright (C) 2018-2021 Intel Corporation
+// SPDX-License-Identifier: Apache-2.0
 //
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
 //*****************************************************************************
 
 #pragma once

diff --git a/ngraph/test/files/paddlepaddle/gen_scripts/generate_pool2d.py b/ngraph/test/files/paddlepaddle/gen_scripts/generate_pool2d.py
@@ -168,7 +168,54 @@ def main():
         }
         # shape of out_6: [2, 4, 3, 3] which is different from out_1
         pool2d(pooling_type+'Pool_test6', data_NHWC, pdpd_attrs)
-        #   
+
+        # example 7:
+        # pool_padding size is 1
+        pdpd_attrs = {
+                    'pool_size':[3,3],
+                    'pool_type' : pooling_type,
+                    'pool_stride' : [3,3],
+                    'pool_padding':2,
+                    'global_pooling':False,
+                    'ceil_mode':False,
+                    'exclusive':True,
+                    'data_format':"NCHW"
+        }
+        pool2d(pooling_type+'Pool_test7', data_NCHW, pdpd_attrs)
+
+        #input data for test8 and test9
+        N_data1, C_data1, H_data1, W_data1 = 2, 3, 8, 8
+        data1 = np.arange(N_data1*C_data1*H_data1*W_data1).astype(data_type)
+        data1_NCHW = data1.reshape(N_data1, C_data1, H_data1, W_data1)
+        data1_NHWC = data1.reshape(N_data1, H_data1, W_data1, C_data1)  
+        # example 8:
+        # pool_padding size is 4: [pad_height_top, pad_height_bottom, pad_width_left, pad_width_right]
+        pdpd_attrs = {
+                    'pool_size':[3,3],
+                    'pool_type' : pooling_type,
+                    'pool_stride' : [3,3],
+                    'pool_padding':[2, 1, 2, 1],
+                    'global_pooling':False,
+                    'ceil_mode':False,
+                    'exclusive':True,
+                    'data_format':"NCHW"
+        }  
+        pool2d(pooling_type+'Pool_test8', data1_NCHW, pdpd_attrs)
+
+        # example 9:
+        # input=data_NCHW and pool_padding is [[0,0], [0,0], [pad_height_top, pad_height_bottom], [pad_width_left, pad_width_right]]
+        pdpd_attrs = {
+                    'pool_size':[3,3],
+                    'pool_type' : pooling_type,
+                    'pool_stride' : [3,3],
+                    'pool_padding':[[0,0], [0,0], [2, 1], [2, 1]],
+                    'global_pooling':False,
+                    'ceil_mode':False,
+                    'exclusive':True,
+                    'data_format':"NCHW"
+        }  
+        pool2d(pooling_type+'Pool_test9', data1_NCHW, pdpd_attrs)
+
 
     # adaptive_pool2d
     for i, pooling_type in enumerate(pooling_types):

diff --git a/ngraph/test/files/paddlepaddle/models/models.csv b/ngraph/test/files/paddlepaddle/models/models.csv
@@ -6,6 +6,9 @@ avgPool_test3,
 avgPool_test4,
 avgPool_test5,
 #avgPool_test6,
+avgPool_test7,
+avgPool_test8,
+avgPool_test9,
 batch_norm,
 bilinear_downsample_false_0,
 bilinear_downsample_false_1,
@@ -53,6 +56,9 @@ maxPool_test3,
 maxPool_test4,
 maxPool_test5,
 #maxPool_test6,
+maxPool_test7,
+maxPool_test8,
+maxPool_test9,
 nearest_downsample_false_0,
 nearest_upsample_false_0,
 pad3d_test1,