experiments/STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Merge pull request from Intel-tensorflow:yang/mirrorpad

Browse Source 
PiperOrigin-RevId: 287920237
Change-Id: I3e7aacbdd52584bfb99cc347ade8c6429d89196d
This commit is contained in:
TensorFlower Gardener 2020-01-02 17:22:31 -08:00
commit 448c0b41bc
4 changed files with 284 additions and 1 deletions

18
tensorflow/core/kernels/BUILD
View File

@ -3166,6 +3166,7 @@ tf_cc_tests(
"adjust_contrast_op_test.cc",
"colorspace_op_test.cc",
"crop_and_resize_op_test.cc",
"mirror_pad_op_test.cc",
"non_max_suppression_op_test.cc",
"resize_area_op_test.cc",
"resize_bicubic_op_test.cc",
@ -3178,6 +3179,7 @@ tf_cc_tests(
}),
deps = [
":image",
":mirror_pad_op",
":ops_testutil",
":ops_util",
":sampling_kernels",
@ -3244,6 +3246,22 @@ tf_cuda_cc_test(
],
)
tf_cuda_cc_test(
name = "mirror_pad_op_benchmark_test",
srcs = ["mirror_pad_op_benchmark_test.cc"],
deps = [
":mirror_pad_op",
":ops_testutil",
":ops_util",
"//tensorflow/core:core_cpu",
"//tensorflow/core:framework",
"//tensorflow/core:protos_all_cc",
"//tensorflow/core:test",
"//tensorflow/core:test_main",
"//tensorflow/core:testlib",
],
)
tf_cuda_cc_test(
name = "non_max_suppression_op_gpu_test",
srcs = ["non_max_suppression_op_gpu_test.cc"],

3
tensorflow/core/kernels/mirror_pad_op.h
View File

@ -223,7 +223,8 @@ struct TensorEvaluator<const TensorMirrorPadOp<PaddingDimensions, ArgType>,
const Index right =
(dimensions_[dim] - padding_[dim].second) * output_strides_[dim];
if (left <= index && (index + kPacketSize - 1) < right) {
const Index index_mod = index % (dimensions_[dim] * output_strides_[dim]);
if (left <= index_mod && (index_mod + kPacketSize - 1) < right) {
return impl_.template packet<Unaligned>(input_index);
}

59
tensorflow/core/kernels/mirror_pad_op_benchmark_test.cc Normal file
View File

@ -0,0 +1,59 @@
/* Copyright 2019 The TensorFlow Authors. All Rights Reserved.
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 "tensorflow/core/common_runtime/kernel_benchmark_testlib.h"
#include "tensorflow/core/framework/tensor.h"
#include "tensorflow/core/graph/node_builder.h"
#include "tensorflow/core/platform/test.h"
#include "tensorflow/core/platform/test_benchmark.h"
namespace tensorflow {
static Graph* BM_MirrorPad(int batches, int height, int width, int depth,
int pad, const char* mode) {
Graph* g = new Graph(OpRegistry::Global());
Tensor in(DT_FLOAT, TensorShape({batches, height, width, depth}));
in.flat<float>().setRandom();
Tensor padding(DT_INT32, TensorShape({4, 2}));
auto boxes_tensor = padding.flat<int>().setZero();
for (int i = 2; i < 6; i++) boxes_tensor(i) = pad;
Node* ret;
TF_CHECK_OK(NodeBuilder(g->NewName("n"), "MirrorPad")
.Input(test::graph::Constant(g, in))
.Input(test::graph::Constant(g, padding))
.Attr("mode", mode)
.Finalize(g, &ret));
return g;
}
#define BM_MirrorPadDev(DEVICE, B, W, H, D, P, MODE) \
static void BM_MirrorPad_##DEVICE##_##B##_##W##_##H##_##D##_##P##_##MODE( \
int iters) { \
testing::ItemsProcessed(iters* B*(W + 2 * P) * (H + 2 * P) * D / 32); \
test::Benchmark(#DEVICE, BM_MirrorPad(B, W, H, D, P, #MODE)).Run(iters); \
} \
BENCHMARK(BM_MirrorPad_##DEVICE##_##B##_##W##_##H##_##D##_##P##_##MODE);
BM_MirrorPadDev(cpu, 1, 16, 16, 32, 1, REFLECT);
BM_MirrorPadDev(cpu, 1, 16, 16, 32, 8, REFLECT);
BM_MirrorPadDev(cpu, 1, 512, 512, 16, 1, REFLECT);
BM_MirrorPadDev(cpu, 1, 512, 512, 16, 256, REFLECT);
BM_MirrorPadDev(cpu, 1, 16, 16, 32, 1, SYMMETRIC);
BM_MirrorPadDev(cpu, 1, 16, 16, 32, 8, SYMMETRIC);
BM_MirrorPadDev(cpu, 1, 512, 512, 16, 1, SYMMETRIC);
BM_MirrorPadDev(cpu, 1, 512, 512, 16, 256, SYMMETRIC);
} // namespace tensorflow

205
tensorflow/core/kernels/mirror_pad_op_test.cc Normal file
View File

@ -0,0 +1,205 @@
/* Copyright 2019 The TensorFlow Authors. All Rights Reserved.
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 "tensorflow/core/framework/allocator.h"
#include "tensorflow/core/framework/fake_input.h"
#include "tensorflow/core/framework/node_def_builder.h"
#include "tensorflow/core/framework/op_kernel.h"
#include "tensorflow/core/framework/register_types.h"
#include "tensorflow/core/framework/tensor.h"
#include "tensorflow/core/framework/tensor_testutil.h"
#include "tensorflow/core/framework/tensor_util.h"
#include "tensorflow/core/framework/types.h"
#include "tensorflow/core/framework/types.pb.h"
#include "tensorflow/core/kernels/ops_testutil.h"
#include "tensorflow/core/kernels/ops_util.h"
#include "tensorflow/core/lib/core/status_test_util.h"
#include "tensorflow/core/lib/strings/str_util.h"
#include "tensorflow/core/platform/test.h"
namespace tensorflow {
class MirrorPadOpTest : public OpsTestBase {
protected:
template <typename T>
void MakeOp(const string& mode) {
TF_EXPECT_OK(NodeDefBuilder("mirror_pad_op", "MirrorPad")
.Input(FakeInput(DataTypeToEnum<T>::value))
.Input(FakeInput(DT_INT32))
.Attr("mode", mode)
.Finalize(node_def()));
TF_EXPECT_OK(InitOp());
}
};
#define REGISTER_TEST(T) \
TEST_F(MirrorPadOpTest, TestMirrorPadReflect##T) { \
MakeOp<T>("REFLECT"); \
AddInputFromArray<T>(TensorShape({1, 2, 3, 1}), {1, 2, 3, 4, 5, 6}); \
AddInputFromArray<int32>(TensorShape({4, 2}), {0, 0, 1, 1, 2, 2, 0, 0}); \
TF_ASSERT_OK(RunOpKernel()); \
\
Tensor expected(allocator(), DataTypeToEnum<T>::value, \
TensorShape({1, 4, 7, 1})); \
test::FillValues<T>(&expected, \
{6, 5, 4, 5, 6, 5, 4, 3, 2, 1, 2, 3, 2, 1, \
6, 5, 4, 5, 6, 5, 4, 3, 2, 1, 2, 3, 2, 1}); \
test::ExpectTensorEqual<T>(expected, *GetOutput(0)); \
} \
\
TEST_F(MirrorPadOpTest, TestMirrorPadSymmetric##T) { \
MakeOp<T>("SYMMETRIC"); \
AddInputFromArray<T>(TensorShape({1, 2, 1, 3}), {1, 2, 3, 4, 5, 6}); \
AddInputFromArray<int32>(TensorShape({4, 2}), {1, 1, 0, 0, 0, 0, 2, 2}); \
TF_ASSERT_OK(RunOpKernel()); \
\
Tensor expected(allocator(), DataTypeToEnum<T>::value, \
TensorShape({3, 2, 1, 7})); \
test::FillValues<T>( \
&expected, \
{2, 1, 1, 2, 3, 3, 2, 5, 4, 4, 5, 6, 6, 5, 2, 1, 1, 2, 3, 3, 2, \
5, 4, 4, 5, 6, 6, 5, 2, 1, 1, 2, 3, 3, 2, 5, 4, 4, 5, 6, 6, 5}); \
test::ExpectTensorEqual<T>(expected, *GetOutput(0)); \
}
REGISTER_TEST(float)
REGISTER_TEST(double)
REGISTER_TEST(uint8)
REGISTER_TEST(uint16)
REGISTER_TEST(int8)
REGISTER_TEST(int16)
REGISTER_TEST(int32)
REGISTER_TEST(int64)
#undef REGISTER_TEST
TEST_F(MirrorPadOpTest, TestMirrorPadReflectLargeInput) {
MakeOp<float>("REFLECT");
// Generate a relatively large input
const int kInput = 1000;
const int kPad = 10;
const int kOutput = kInput + 2 * kPad;
// Input:
// 0, 1, 2, ..., 999
// 0, 1, 2, ..., 999
// ... (altogether 1000 lines)
// 0, 1, 2, ..., 999
AddInput<float>(TensorShape({1, kInput, kInput, 1}),
[](int i) -> float { return i % kInput; });
AddInputFromArray<int32>(TensorShape({4, 2}),
{0, 0, kPad, kPad, kPad, kPad, 0, 0});
TF_ASSERT_OK(RunOpKernel());
Tensor expected(allocator(), DT_FLOAT, TensorShape({1, kOutput, kOutput, 1}));
test::FillFn<float>(&expected, [](int i) -> float {
i = i % kOutput;
if (0 <= i && i < kPad)
return kPad - i;
else if (kPad <= i && i < kInput + kPad)
return i - kPad;
else if (kInput + kPad <= i && i < kOutput)
return 2 * kInput + kPad - 2 - i;
else
return -1;
});
test::ExpectTensorEqual<float>(expected, *GetOutput(0));
}
TEST_F(MirrorPadOpTest, TestMirrorPadSymmetricLargeInput) {
MakeOp<float>("SYMMETRIC");
// Generate a relatively large input
const int kInput = 1000;
const int kPad = 10;
const int kOutput = kInput + 2 * kPad;
// Input:
// 0, 1, 2, ..., 999
// 0, 1, 2, ..., 999
// ... (altogether 1000 lines)
// 0, 1, 2, ..., 999
AddInput<float>(TensorShape({1, kInput, kInput, 1}),
[](int i) -> float { return i % kInput; });
AddInputFromArray<int32>(TensorShape({4, 2}),
{0, 0, kPad, kPad, kPad, kPad, 0, 0});
TF_ASSERT_OK(RunOpKernel());
Tensor expected(allocator(), DT_FLOAT, TensorShape({1, kOutput, kOutput, 1}));
test::FillFn<float>(&expected, [](int i) -> float {
i = i % kOutput;
if (0 <= i && i < kPad)
return kPad - i - 1;
else if (kPad <= i && i < kInput + kPad)
return i - kPad;
else if (kInput + kPad <= i && i < kOutput)
return 2 * kInput + kPad - 1 - i;
else
return -1;
});
test::ExpectTensorEqual<float>(expected, *GetOutput(0));
}
class MirrorPadGradOpTest : public OpsTestBase {
protected:
template <typename T>
void MakeOp(const string& mode) {
TF_EXPECT_OK(NodeDefBuilder("mirror_pad_grad_op", "MirrorPadGrad")
.Input(FakeInput(DataTypeToEnum<T>::value))
.Input(FakeInput(DT_INT32))
.Attr("mode", mode)
.Finalize(node_def()));
TF_EXPECT_OK(InitOp());
}
};
#define REGISTER_TEST(T) \
TEST_F(MirrorPadGradOpTest, TestMirrorPadGradReflect##T) { \
MakeOp<T>("REFLECT"); \
AddInput<T>(TensorShape({1, 4, 7, 1}), [](int i) -> T { return i % 7; }); \
AddInputFromArray<int32>(TensorShape({4, 2}), {0, 0, 1, 1, 2, 2, 0, 0}); \
TF_ASSERT_OK(RunOpKernel()); \
\
Tensor expected(allocator(), DataTypeToEnum<T>::value, \
TensorShape({1, 2, 3, 1})); \
test::FillValues<T>(&expected, {16, 18, 8, 16, 18, 8}); \
test::ExpectTensorEqual<T>(expected, *GetOutput(0)); \
} \
\
TEST_F(MirrorPadGradOpTest, TestMirrorPadGradSymmetric##T) { \
MakeOp<T>("SYMMETRIC"); \
AddInput<T>(TensorShape({3, 2, 1, 7}), [](int i) -> T { return i % 7; }); \
AddInputFromArray<int32>(TensorShape({4, 2}), {1, 1, 0, 0, 0, 0, 2, 2}); \
TF_ASSERT_OK(RunOpKernel()); \
\
Tensor expected(allocator(), DataTypeToEnum<T>::value, \
TensorShape({1, 2, 1, 3})); \
test::FillValues<T>(&expected, {9, 27, 27, 9, 27, 27}); \
test::ExpectTensorEqual<T>(expected, *GetOutput(0)); \
}
REGISTER_TEST(float)
REGISTER_TEST(double)
REGISTER_TEST(uint8)
REGISTER_TEST(uint16)
REGISTER_TEST(int8)
REGISTER_TEST(int16)
REGISTER_TEST(int32)
REGISTER_TEST(int64)
#undef REGISTER_TEST
} // namespace tensorflow