mindspore-Ecosystem
/
mindspore
21
0
Fork 27
Code Issues Pull Requests 6 Projects Releases Wiki Activity

[feat][assistant][I4S2FM]expand the normalization operation of high dimension

This commit is contained in:
lianlinghang 2022-08-26 17:00:23 +08:00
parent d6e6872183
commit d443cee406
5 changed files with 139 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

35
mindspore/ccsrc/minddata/dataset/kernels/image/normalize_op.cc
View File

@ -18,6 +18,7 @@
#include <random>
#include <vector>
#include "minddata/dataset/kernels/data/data_utils.h"
#ifndef ENABLE_ANDROID
#include "minddata/dataset/kernels/image/image_utils.h"
#else
@ -33,11 +34,41 @@ NormalizeOp::NormalizeOp(const std::vector<float> &mean, const std::vector<float
Status NormalizeOp::Compute(const std::shared_ptr<Tensor> &input, std::shared_ptr<Tensor> *output) {
IO_CHECK(input, output);
// Doing the Normalization
auto input_shape = input->shape();
dsize_t rank = input_shape.Rank();
if (rank < kMinImageRank) {
std::string err_msg = "Normalize: input tensor should have at least 2 dimensions, but got: " + std::to_string(rank);
RETURN_STATUS_UNEXPECTED(err_msg);
} else if (rank <= kDefaultImageRank) {
// [H, W] or [H, W, C]
#ifndef ENABLE_ANDROID
return Normalize(input, output, mean_, std_, is_hwc_);
return Normalize(input, output, mean_, std_, is_hwc_);
#else
return Normalize(input, output, mean_, std_);
return Normalize(input, output, mean_, std_);
#endif
} else {
// reshape [..., H, W, C] to [N, H, W, C]
dsize_t num_batch = input->Size() / (input_shape[-3] * input_shape[-2] * input_shape[-1]);
TensorShape new_shape({num_batch, input_shape[-3], input_shape[-2], input_shape[-1]});
RETURN_IF_NOT_OK(input->Reshape(new_shape));
// split [N, H, W, C] to N [H, W, C], and normalize N [H, W, C]
std::vector<std::shared_ptr<Tensor>> input_vector_hwc, output_vector_hwc;
RETURN_IF_NOT_OK(BatchTensorToTensorVector(input, &input_vector_hwc));
for (auto input_hwc : input_vector_hwc) {
std::shared_ptr<Tensor> normalize;
#ifndef ENABLE_ANDROID
RETURN_IF_NOT_OK(Normalize(input_hwc, &normalize, mean_, std_, is_hwc_));
#else
RETURN_IF_NOT_OK(Normalize(input_hwc, &normalize, mean_, std_));
#endif
output_vector_hwc.push_back(normalize);
}
// integrate N [H, W, C] to [N, H, W, C], and reshape [..., H, W, C]
RETURN_IF_NOT_OK(TensorVectorToBatchTensor(output_vector_hwc, &(*output)));
RETURN_IF_NOT_OK((*output)->Reshape(input_shape));
return Status::OK();
}
}
void NormalizeOp::Print(std::ostream &out) const {

4
mindspore/python/mindspore/dataset/vision/c_transforms.py
View File

@ -819,8 +819,8 @@ class Normalize(ImageTensorOperation):
TypeError: If `mean` is not of type sequence.
TypeError: If `std` is not of type sequence.
ValueError: If `mean` is not in range [0.0, 255.0].
ValueError: If `mean` is not in range (0.0, 255.0].
RuntimeError: If given tensor shape is not <H, W> or <H, W, C>.
ValueError: If `std` is not in range (0.0, 255.0].
RuntimeError: If given tensor shape is not <H, W> or <...,H, W, C>.
Supported Platforms:
``CPU`` ``Ascend`` ``GPU``

3
mindspore/python/mindspore/dataset/vision/transforms.py
View File

@ -1401,7 +1401,8 @@ class Normalize(ImageTensorOperation):
TypeError: If `std` is not of type sequence.
TypeError: If `is_hwc` is not of type bool.
ValueError: If `mean` is not in range [0.0, 255.0].
ValueError: If `mean` is not in range (0.0, 255.0].
ValueError: If `std` is not in range (0.0, 255.0].
RuntimeError: If given tensor shape is not <H, W> or <...,H, W, C>.
Supported Platforms:
``CPU``

44
tests/ut/cpp/dataset/normalize_op_test.cc
View File

@ -15,6 +15,7 @@
*/
#include "common/common.h"
#include "common/cvop_common.h"
#include "minddata/dataset/kernels/data/data_utils.h"
#include "minddata/dataset/kernels/image/normalize_op.h"
#include "minddata/dataset/core/cv_tensor.h"
#include "utils/log_adapter.h"
@ -27,9 +28,9 @@ class MindDataTestNormalizeOP : public UT::CVOP::CVOpCommon {
MindDataTestNormalizeOP() : CVOpCommon() {}
};
/// Feature: Normalize op
/// Description: Test input parameters at Runtime level
/// Expectation: Results are successfully outputted.
/// Feature: Normalize
/// Description: Normalize the image and save
/// Expectation: normalized image saves successfully
TEST_F(MindDataTestNormalizeOP, TestOp) {
MS_LOG(INFO) << "Doing TestNormalizeOp::TestOp2.";
std::shared_ptr<Tensor> output_tensor;
@ -55,3 +56,40 @@ TEST_F(MindDataTestNormalizeOP, TestOp) {
cv::FileStorage file(output_filename, cv::FileStorage::WRITE);
file << "imageData" << cv_output_image;
}
/// Feature: Normalize
/// Description: Test Normalize with 4 dimension tensor
/// Expectation: The result is as expected
TEST_F(MindDataTestNormalizeOP, TestOp4Dim) {
MS_LOG(INFO) << "Doing TestNormalizeOp-TestOp4Dim.";
std::shared_ptr<Tensor> output_tensor;
// construct a fake 4 dimension data
std::shared_ptr<Tensor> input_tensor_cp;
ASSERT_OK(Tensor::CreateFromTensor(input_tensor_, &input_tensor_cp));
std::vector<std::shared_ptr<Tensor>> tensor_list;
tensor_list.push_back(input_tensor_cp);
tensor_list.push_back(input_tensor_cp);
TensorShape shape = input_tensor_cp->shape();
std::shared_ptr<Tensor> input_4d;
ASSERT_OK(TensorVectorToBatchTensor(tensor_list, &input_4d));
std::vector<float> mean = {121.0, 115.0, 100.0};
std::vector<float> std = {70.0, 68.0, 71.0};
// Normalize Op
std::unique_ptr<NormalizeOp> op = std::make_unique<NormalizeOp>(mean, std, true);
EXPECT_TRUE(op->OneToOne());
Status s = op->Compute(input_4d, &output_tensor);
EXPECT_TRUE(s.IsOk());
std::string output_filename = GetFilename();
output_filename.replace(output_filename.end() - 8, output_filename.end(), "imagefolder/normalizeOpVideoOut.yml");
std::shared_ptr<CVTensor> p = CVTensor::AsCVTensor(output_tensor);
cv::Mat cv_output_video;
cv_output_video = p->mat();
MS_LOG(DEBUG) << "Storing output file to : " << output_filename << std::endl;
cv::FileStorage file(output_filename, cv::FileStorage::WRITE);
file << "videoData" << cv_output_video;
}

61
tests/ut/python/dataset/test_normalize_op.py
View File

@ -160,6 +160,65 @@ def test_normalize_op_chw(plot=False):
num_iter += 1
def test_normalize_op_video():
"""
Feature: Normalize op
Description: Test NormalizeOp in Cpp transformation with 4 dimension input,
where the input tensor is (..., T, H, W, C)
Expectation: The dataset is processed successfully
"""
logger.info("Test NormalizeOp in cpp transformations with 4 dimension input")
mean = [121.0, 115.0, 100.0]
std = [70.0, 68.0, 71.0]
input_np_original = np.array([[87, 88, 232, 239],
[11, 229, 22, 79],
[250, 20, 173, 213]], dtype=np.float32)
expect_output = np.array([[-1.0714285, -1.3088236, 0.11267605, 1.7142857],
[-0.35211268, -0.55714285, 1.0735294, 0.52112675],
[-0.27941176, 0.43661973, 1.3428571, -0.9411765]], dtype=np.float32)
shape = (2, 2, 1, 3)
input_np_original = input_np_original.reshape(shape)
expect_output = expect_output.reshape(shape)
# define operations
normalize_op = vision.Normalize(mean, std, True)
# doing the Normalization
vidio_de_normalized = normalize_op(input_np_original)
mse = diff_mse(vidio_de_normalized, expect_output)
assert mse < 0.01
def test_normalize_op_5d():
"""
Feature: Normalize op
Description: Test NormalizeOp in Cpp transformation with 5 dim input, where the input tensor is (..., T, H, W, C)
Expectation: The dataset is processed successfully
"""
logger.info("Test NormalizeOp in cpp transformations with 5 dimension input")
mean = [121.0, 115.0, 100.0]
std = [70.0, 68.0, 71.0]
input_np_original = np.array([[87, 88, 232, 239],
[11, 229, 22, 79],
[250, 20, 173, 213]], dtype=np.float32)
expect_output = np.array([[-1.0714285, -1.3088236, 0.11267605, 1.7142857],
[-0.35211268, -0.55714285, 1.0735294, 0.52112675],
[-0.27941176, 0.43661973, 1.3428571, -0.9411765]], dtype=np.float32)
shape = (2, 1, 2, 1, 3)
input_np_original = input_np_original.reshape(shape)
expect_output = expect_output.reshape(shape)
# define operations
normalize_op = vision.Normalize(mean, std, True)
# doing the Normalization
vidio_de_normalized = normalize_op(input_np_original)
mse = diff_mse(vidio_de_normalized, expect_output)
assert mse < 0.01
def test_decode_op():
"""
Feature: Decode op
@ -466,6 +525,8 @@ if __name__ == "__main__":
test_decode_normalize_op()
test_normalize_op_hwc(plot=True)
test_normalize_op_chw(plot=True)
test_normalize_op_video()
test_normalize_op_5d()
test_normalize_md5_01()
test_normalize_md5_02()
test_normalize_exception_unequal_size_1()