forked from mindspore-Ecosystem/mindspore
469 lines
18 KiB
Python
469 lines
18 KiB
Python
# Copyright 2019-2022 Huawei Technologies Co., Ltd
|
|
#
|
|
# 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.
|
|
# ==============================================================================
|
|
"""
|
|
Testing Normalize op in DE
|
|
"""
|
|
import numpy as np
|
|
from PIL import Image
|
|
import mindspore.dataset as ds
|
|
import mindspore.dataset.transforms.transforms
|
|
import mindspore.dataset.vision.transforms as vision
|
|
from mindspore import log as logger
|
|
from util import diff_mse, save_and_check_md5, visualize_image
|
|
|
|
DATA_DIR = ["../data/dataset/test_tf_file_3_images/train-0000-of-0001.data"]
|
|
SCHEMA_DIR = "../data/dataset/test_tf_file_3_images/datasetSchema.json"
|
|
|
|
GENERATE_GOLDEN = False
|
|
|
|
|
|
def normalize_np(image, mean, std):
|
|
"""
|
|
Apply the Normalization
|
|
"""
|
|
# DE decodes the image in RGB by default, hence
|
|
# the values here are in RGB
|
|
image = np.array(image, np.float32)
|
|
image = image - np.array(mean)
|
|
image = image * (1.0 / np.array(std))
|
|
return image
|
|
|
|
|
|
def util_test_normalize(mean, std, add_to_pil):
|
|
"""
|
|
Utility function for testing Normalize. Input arguments are given by other tests
|
|
"""
|
|
if not add_to_pil:
|
|
# define map operations
|
|
decode_op = vision.Decode()
|
|
normalize_op = vision.Normalize(mean, std, True)
|
|
# Generate dataset
|
|
data = ds.TFRecordDataset(DATA_DIR, SCHEMA_DIR, columns_list=["image"], shuffle=False)
|
|
data = data.map(operations=decode_op, input_columns=["image"])
|
|
data = data.map(operations=normalize_op, input_columns=["image"])
|
|
else:
|
|
# define map operations
|
|
transforms = [
|
|
vision.Decode(True),
|
|
vision.ToTensor(),
|
|
vision.Normalize(mean, std, False)
|
|
]
|
|
transform = mindspore.dataset.transforms.transforms.Compose(transforms)
|
|
# Generate dataset
|
|
data = ds.TFRecordDataset(DATA_DIR, SCHEMA_DIR, columns_list=["image"], shuffle=False)
|
|
data = data.map(operations=transform, input_columns=["image"])
|
|
return data
|
|
|
|
|
|
def util_test_normalize_grayscale(num_output_channels, mean, std):
|
|
"""
|
|
Utility function for testing Normalize. Input arguments are given by other tests
|
|
"""
|
|
transforms = [
|
|
vision.Decode(True),
|
|
vision.Grayscale(num_output_channels),
|
|
vision.ToTensor(),
|
|
vision.Normalize(mean, std, False)
|
|
]
|
|
transform = mindspore.dataset.transforms.transforms.Compose(transforms)
|
|
# Generate dataset
|
|
data = ds.TFRecordDataset(DATA_DIR, SCHEMA_DIR, columns_list=["image"], shuffle=False)
|
|
data = data.map(operations=transform, input_columns=["image"])
|
|
return data
|
|
|
|
|
|
def test_normalize_op_hwc(plot=False):
|
|
"""
|
|
Feature: Normalize op
|
|
Description: Test Normalize with Decode versus NumPy comparison
|
|
Expectation: Test succeeds. MSE difference is negligible.
|
|
"""
|
|
logger.info("Test Normalize in with hwc")
|
|
mean = [121.0, 115.0, 100.0]
|
|
std = [70.0, 68.0, 71.0]
|
|
# define map operations
|
|
decode_op = vision.Decode()
|
|
normalize_op = vision.Normalize(mean, std, True)
|
|
|
|
# First dataset
|
|
data1 = ds.TFRecordDataset(DATA_DIR, SCHEMA_DIR, columns_list=["image"], shuffle=False)
|
|
data1 = data1.map(operations=decode_op, input_columns=["image"])
|
|
data1 = data1.map(operations=normalize_op, input_columns=["image"])
|
|
|
|
# Second dataset
|
|
data2 = ds.TFRecordDataset(DATA_DIR, SCHEMA_DIR, columns_list=["image"], shuffle=False)
|
|
data2 = data2.map(operations=decode_op, input_columns=["image"])
|
|
|
|
num_iter = 0
|
|
for item1, item2 in zip(data1.create_dict_iterator(num_epochs=1, output_numpy=True),
|
|
data2.create_dict_iterator(num_epochs=1, output_numpy=True)):
|
|
image_de_normalized = item1["image"]
|
|
image_original = item2["image"]
|
|
image_np_normalized = normalize_np(image_original, mean, std)
|
|
np.testing.assert_almost_equal(image_de_normalized, image_np_normalized, 2)
|
|
mse = diff_mse(image_de_normalized, image_np_normalized)
|
|
logger.info("image_{}, mse: {}".format(num_iter + 1, mse))
|
|
if plot:
|
|
visualize_image(image_original, image_de_normalized, mse, image_np_normalized)
|
|
num_iter += 1
|
|
|
|
|
|
def test_normalize_op_chw(plot=False):
|
|
"""
|
|
Feature: Normalize op
|
|
Description: Test Normalize with CHW input, Decode(to_pil=True) & ToTensor versus NumPy comparison
|
|
Expectation: Test succeeds. MSE difference is negligible.
|
|
"""
|
|
logger.info("Test Normalize with chw")
|
|
mean = [0.475, 0.45, 0.392]
|
|
std = [0.275, 0.267, 0.278]
|
|
# define map operations
|
|
transforms = [
|
|
vision.Decode(True),
|
|
vision.ToTensor()
|
|
]
|
|
transform = mindspore.dataset.transforms.transforms.Compose(transforms)
|
|
normalize_op = vision.Normalize(mean, std, False)
|
|
|
|
# First dataset
|
|
data1 = ds.TFRecordDataset(DATA_DIR, SCHEMA_DIR, columns_list=["image"], shuffle=False)
|
|
data1 = data1.map(operations=transform, input_columns=["image"])
|
|
data1 = data1.map(operations=normalize_op, input_columns=["image"])
|
|
|
|
# Second dataset
|
|
data2 = ds.TFRecordDataset(DATA_DIR, SCHEMA_DIR, columns_list=["image"], shuffle=False)
|
|
data2 = data2.map(operations=transform, input_columns=["image"])
|
|
|
|
num_iter = 0
|
|
for item1, item2 in zip(data1.create_dict_iterator(num_epochs=1, output_numpy=True),
|
|
data2.create_dict_iterator(num_epochs=1, output_numpy=True)):
|
|
image_de_normalized = (item1["image"].transpose(1, 2, 0) * 255).astype(np.uint8)
|
|
image_np_normalized = (normalize_np(item2["image"].transpose(1, 2, 0), mean, std) * 255).astype(np.uint8)
|
|
mse = diff_mse(image_de_normalized, image_np_normalized)
|
|
logger.info("image_{}, mse: {}".format(num_iter + 1, mse))
|
|
assert mse < 0.01
|
|
if plot:
|
|
image_original = (item2["image"].transpose(1, 2, 0) * 255).astype(np.uint8)
|
|
visualize_image(image_original, image_de_normalized, mse, image_np_normalized)
|
|
num_iter += 1
|
|
|
|
|
|
def test_decode_op():
|
|
"""
|
|
Test Decode op
|
|
"""
|
|
logger.info("Test Decode")
|
|
|
|
data1 = ds.TFRecordDataset(DATA_DIR, SCHEMA_DIR, columns_list=["image", "label"], num_parallel_workers=1,
|
|
shuffle=False)
|
|
|
|
# define map operations
|
|
decode_op = vision.Decode()
|
|
|
|
# apply map operations on images
|
|
data1 = data1.map(operations=decode_op, input_columns=["image"])
|
|
|
|
num_iter = 0
|
|
for item in data1.create_dict_iterator(num_epochs=1):
|
|
logger.info("Looping inside iterator {}".format(num_iter))
|
|
_ = item["image"]
|
|
num_iter += 1
|
|
|
|
|
|
def test_decode_normalize_op():
|
|
"""
|
|
Test Decode op followed by Normalize op
|
|
"""
|
|
logger.info("Test [Decode, Normalize] in one Map")
|
|
|
|
data1 = ds.TFRecordDataset(DATA_DIR, SCHEMA_DIR, columns_list=["image", "label"], num_parallel_workers=1,
|
|
shuffle=False)
|
|
|
|
# define map operations
|
|
decode_op = vision.Decode()
|
|
normalize_op = vision.Normalize([121.0, 115.0, 100.0], [70.0, 68.0, 71.0], True)
|
|
|
|
# apply map operations on images
|
|
data1 = data1.map(operations=[decode_op, normalize_op], input_columns=["image"])
|
|
|
|
num_iter = 0
|
|
for item in data1.create_dict_iterator(num_epochs=1):
|
|
logger.info("Looping inside iterator {}".format(num_iter))
|
|
_ = item["image"]
|
|
num_iter += 1
|
|
|
|
|
|
def test_normalize_md5_01():
|
|
"""
|
|
Test Normalize with md5 check: valid mean and std
|
|
expected to pass
|
|
"""
|
|
logger.info("test_normalize_md5_01")
|
|
data_c = util_test_normalize([121.0, 115.0, 100.0], [70.0, 68.0, 71.0], False)
|
|
data_py = util_test_normalize([0.475, 0.45, 0.392], [0.275, 0.267, 0.278], True)
|
|
|
|
# check results with md5 comparison
|
|
filename1 = "normalize_01_c_result.npz"
|
|
filename2 = "normalize_01_to_pil_result.npz"
|
|
save_and_check_md5(data_c, filename1, generate_golden=GENERATE_GOLDEN)
|
|
save_and_check_md5(data_py, filename2, generate_golden=GENERATE_GOLDEN)
|
|
|
|
|
|
def test_normalize_md5_02():
|
|
"""
|
|
Test Normalize with md5 check: len(mean)=len(std)=1 with RGB images
|
|
expected to pass
|
|
"""
|
|
logger.info("test_normalize_md5_02")
|
|
data_py = util_test_normalize([0.475], [0.275], True)
|
|
|
|
# check results with md5 comparison
|
|
filename2 = "normalize_02_to_pil_result.npz"
|
|
save_and_check_md5(data_py, filename2, generate_golden=GENERATE_GOLDEN)
|
|
|
|
|
|
def test_normalize_exception_unequal_size_1():
|
|
"""
|
|
Feature: Normalize op
|
|
Description: Test Normalize with error input: len(mean) != len(std)
|
|
Expectation: ValueError raised
|
|
"""
|
|
logger.info("test_normalize_exception_unequal_size_1")
|
|
try:
|
|
_ = vision.Normalize([100, 250, 125], [50, 50, 75, 75])
|
|
except ValueError as e:
|
|
logger.info("Got an exception in DE: {}".format(str(e)))
|
|
assert str(e) == "Length of mean and std must be equal."
|
|
|
|
|
|
def test_normalize_exception_out_of_range():
|
|
"""
|
|
Feature: Normalize op
|
|
Description: Test Normalize with error input: mean, std out of range
|
|
Expectation: ValueError raised
|
|
"""
|
|
logger.info("test_normalize_exception_out_of_range")
|
|
try:
|
|
_ = vision.Normalize([256, 250, 125], [50, 75, 75])
|
|
except ValueError as e:
|
|
logger.info("Got an exception in DE: {}".format(str(e)))
|
|
assert "not within the required interval" in str(e)
|
|
try:
|
|
_ = vision.Normalize([255, 250, 125], [0, 75, 75])
|
|
except ValueError as e:
|
|
logger.info("Got an exception in DE: {}".format(str(e)))
|
|
assert "not within the required interval" in str(e)
|
|
|
|
|
|
def test_normalize_exception_unequal_size_2():
|
|
"""
|
|
Feature: Normalize op
|
|
Description: Test Normalize with error input: len(mean) != len(std)
|
|
Expectation: ValueError raised
|
|
"""
|
|
logger.info("test_normalize_exception_unequal_size_2")
|
|
try:
|
|
_ = vision.Normalize([0.50, 0.30, 0.75], [0.18, 0.32, 0.71, 0.72], False)
|
|
except ValueError as e:
|
|
logger.info("Got an exception in DE: {}".format(str(e)))
|
|
assert str(e) == "Length of mean and std must be equal."
|
|
|
|
|
|
def test_normalize_exception_invalid_size():
|
|
"""
|
|
Feature: Normalize op
|
|
Description: Test Normalize with error input: len(mean)=len(std)=2
|
|
Expectation: RuntimeError raised
|
|
"""
|
|
logger.info("test_normalize_exception_invalid_size")
|
|
data = util_test_normalize([0.75, 0.25], [0.18, 0.32], False)
|
|
try:
|
|
_ = data.create_dict_iterator(num_epochs=1).__next__()
|
|
except RuntimeError as e:
|
|
logger.info("Got an exception in DE: {}".format(str(e)))
|
|
assert "Normalize: number of channels does not match the size of mean and std vectors" in str(e)
|
|
|
|
|
|
def test_normalize_exception_invalid_range():
|
|
"""
|
|
Feature: Normalize op
|
|
Description: Test Normalize with error input: value is not in range [0,1]
|
|
Expectation: ValueError raised
|
|
"""
|
|
logger.info("test_normalize_exception_invalid_range")
|
|
try:
|
|
_ = vision.Normalize([0.75, 1.25, 0.5], [0.1, 0.18, 1.32], False)
|
|
except ValueError as e:
|
|
logger.info("Got an exception in DE: {}".format(str(e)))
|
|
assert "Input mean_value is not within the required interval of [0.0, 1.0]." in str(e)
|
|
|
|
|
|
def test_normalize_grayscale_md5_01():
|
|
"""
|
|
Test Normalize with md5 check: len(mean)=len(std)=1 with 1 channel grayscale images
|
|
expected to pass
|
|
"""
|
|
logger.info("test_normalize_grayscale_md5_01")
|
|
data = util_test_normalize_grayscale(1, [0.5], [0.175])
|
|
# check results with md5 comparison
|
|
filename = "normalize_03_to_pil_result.npz"
|
|
save_and_check_md5(data, filename, generate_golden=GENERATE_GOLDEN)
|
|
|
|
|
|
def test_normalize_grayscale_md5_02():
|
|
"""
|
|
Test Normalize with md5 check: len(mean)=len(std)=3 with 3 channel grayscale images
|
|
expected to pass
|
|
"""
|
|
logger.info("test_normalize_grayscale_md5_02")
|
|
data = util_test_normalize_grayscale(3, [0.5, 0.5, 0.5], [0.175, 0.235, 0.512])
|
|
# check results with md5 comparison
|
|
filename = "normalize_04_to_pil_result.npz"
|
|
save_and_check_md5(data, filename, generate_golden=GENERATE_GOLDEN)
|
|
|
|
|
|
def test_normalize_grayscale_exception():
|
|
"""
|
|
Test Normalize: len(mean)=len(std)=3 with 1 channel grayscale images
|
|
expected to raise RuntimeError
|
|
"""
|
|
logger.info("test_normalize_grayscale_exception")
|
|
try:
|
|
_ = util_test_normalize_grayscale(1, [0.5, 0.5, 0.5], [0.175, 0.235, 0.512])
|
|
except RuntimeError as e:
|
|
logger.info("Got an exception in DE: {}".format(str(e)))
|
|
assert "Input is not within the required range" in str(e)
|
|
|
|
|
|
def test_multiple_channels():
|
|
logger.info("test_multiple_channels")
|
|
|
|
def util_test(item, mean, std):
|
|
data = ds.NumpySlicesDataset([item], shuffle=False)
|
|
data = data.map(vision.Normalize(mean, std, True))
|
|
for d in data.create_tuple_iterator(num_epochs=1, output_numpy=True):
|
|
actual = d[0]
|
|
mean = np.array(mean, dtype=item.dtype)
|
|
std = np.array(std, dtype=item.dtype)
|
|
expected = item
|
|
if len(item.shape) != 1 and len(mean) == 1:
|
|
mean = [mean[0]] * expected.shape[-1]
|
|
std = [std[0]] * expected.shape[-1]
|
|
if len(item.shape) == 2:
|
|
expected = np.expand_dims(expected, 2)
|
|
for c in range(expected.shape[-1]):
|
|
expected[:, :, c] = (expected[:, :, c] - mean[c]) / std[c]
|
|
expected = expected.squeeze()
|
|
|
|
np.testing.assert_almost_equal(actual, expected, decimal=6)
|
|
|
|
util_test(np.ones(shape=[2, 2, 3]), mean=[0.5, 0.6, 0.7], std=[0.1, 0.2, 0.3])
|
|
util_test(np.ones(shape=[20, 45, 3]) * 1.3, mean=[0.5, 0.6, 0.7], std=[0.1, 0.2, 0.3])
|
|
util_test(np.ones(shape=[20, 45, 4]) * 1.3, mean=[0.5, 0.6, 0.7, 0.8], std=[0.1, 0.2, 0.3, 0.4])
|
|
util_test(np.ones(shape=[2, 2]), mean=[0.5], std=[0.1])
|
|
util_test(np.ones(shape=[2, 2, 5]), mean=[0.5], std=[0.1])
|
|
util_test(np.ones(shape=[6, 6, 129]), mean=[0.5] * 129, std=[0.1] * 129)
|
|
util_test(np.ones(shape=[6, 6, 129]), mean=[0.5], std=[0.1])
|
|
|
|
|
|
def test_normalize_eager_hwc():
|
|
"""
|
|
Feature: Normalize op
|
|
Description: Test eager support for Normalize C implementation with HWC input
|
|
Expectation: Receive non-None output image from op
|
|
"""
|
|
img_in = Image.open("../data/dataset/apple.jpg").convert("RGB")
|
|
mean_vec = [1, 100, 255]
|
|
std_vec = [1, 20, 255]
|
|
normalize_op = vision.Normalize(mean=mean_vec, std=std_vec)
|
|
img_out = normalize_op(img_in)
|
|
assert img_out is not None
|
|
|
|
|
|
def test_normalize_eager_chw():
|
|
"""
|
|
Feature: Normalize op
|
|
Description: Test eager support for Normalize C implementation with CHW input
|
|
Expectation: Receive non-None output image from op
|
|
"""
|
|
img_in = Image.open("../data/dataset/apple.jpg").convert("RGB")
|
|
img_in = vision.ToTensor()(img_in)
|
|
mean_vec = [0.1, 0.5, 1.0]
|
|
std_vec = [0.1, 0.4, 1.0]
|
|
normalize_op = vision.Normalize(mean=mean_vec, std=std_vec, is_hwc=False)
|
|
img_out = normalize_op(img_in)
|
|
assert img_out is not None
|
|
|
|
|
|
def test_normalize_op_comp_chw():
|
|
"""
|
|
Feature: Normalize op
|
|
Description: Test Normalize with CHW input, Decode(to_pil=True) & ToTensor versus Decode(to_pil=False) & HWC2CHW
|
|
comparison.
|
|
Expectation: Test succeeds. MSE difference is negligible.
|
|
"""
|
|
logger.info("Test Normalize with CHW input")
|
|
mean = [0.475, 0.45, 0.392]
|
|
std = [0.275, 0.267, 0.278]
|
|
# define map operations
|
|
transforms = [
|
|
vision.Decode(True),
|
|
vision.ToTensor()
|
|
]
|
|
transform = mindspore.dataset.transforms.transforms.Compose(transforms)
|
|
normalize_op = vision.Normalize(mean, std, False)
|
|
|
|
# First dataset
|
|
data1 = ds.TFRecordDataset(DATA_DIR, SCHEMA_DIR, columns_list=["image"], shuffle=False)
|
|
data1 = data1.map(operations=transform, input_columns=["image"])
|
|
data1 = data1.map(operations=normalize_op, input_columns=["image"])
|
|
|
|
# Second dataset
|
|
data2 = ds.TFRecordDataset(DATA_DIR, SCHEMA_DIR, columns_list=["image"], shuffle=False)
|
|
data2 = data2.map(operations=vision.Decode(), input_columns=["image"])
|
|
data2 = data2.map(operations=vision.HWC2CHW(), input_columns=["image"])
|
|
data2 = data2.map(operations=vision.Normalize(mean, std, False), input_columns=["image"])
|
|
|
|
num_iter = 0
|
|
for item1, item2 in zip(data1.create_dict_iterator(num_epochs=1, output_numpy=True),
|
|
data2.create_dict_iterator(num_epochs=1, output_numpy=True)):
|
|
image_de_normalized = item1["image"]
|
|
image_np_normalized = item2["image"] / 255
|
|
mse = diff_mse(image_de_normalized, image_np_normalized)
|
|
logger.info("image_{}, mse: {}".format(num_iter + 1, mse))
|
|
assert mse < 0.01
|
|
num_iter += 1
|
|
|
|
|
|
if __name__ == "__main__":
|
|
test_decode_op()
|
|
test_decode_normalize_op()
|
|
test_normalize_op_hwc(plot=True)
|
|
test_normalize_op_chw(plot=True)
|
|
test_normalize_md5_01()
|
|
test_normalize_md5_02()
|
|
test_normalize_exception_unequal_size_1()
|
|
test_normalize_exception_out_of_range()
|
|
test_normalize_exception_unequal_size_2()
|
|
test_normalize_exception_invalid_size()
|
|
test_normalize_exception_invalid_range()
|
|
test_normalize_grayscale_md5_01()
|
|
test_normalize_grayscale_md5_02()
|
|
test_normalize_grayscale_exception()
|
|
test_multiple_channels()
|
|
test_normalize_eager_hwc()
|
|
test_normalize_eager_chw()
|
|
test_normalize_op_comp_chw()
|