全局对比度标准化在 python 上给出异常
Global Contrast Normalization giving exception on python
我正在尝试将 GCN 应用于图像列表，但我一直收到此异常，
Saving training image : 0
Traceback (most recent call last):
File "A:/Code/MeusProjetosPython/KaggleMnistProject/Pre_Processing/Main.py", line 12, in <module>
applyGCN(True)
File "A:\Code\MeusProjetosPython\KaggleMnistProject\Pre_Processing\Global_Contrast_Normalization.py", line 123, in applyGCN
mpimg.imsave(trainingFolder + "GCN_Gray_TrainImage_" + str(x), image_aux, cmap=plt.get_cmap('gray'))
File "C:\SciSoft\WinPython-64bit-2.7.9.4\python-2.7.9.amd64\lib\site-packages\matplotlib\image.py", line 1310, in imsave
figsize = [x / float(dpi) for x in (arr.shape[1], arr.shape[0])]
IndexError: tuple index out of range
由于注释，完整代码有点大。但它只是两种方法。
不管怎样，我把代码粘贴在这里。
import numpy
import matplotlib.pyplot as plt
import matplotlib.image as mpimg
import numpy as np
def global_contrast_normalize(X, scale=1., subtract_mean=True, use_std=False, sqrt_bias=0., min_divisor=1e-8):
"""
Global contrast normalizes by (optionally) subtracting the mean
across features and then normalizes by either the vector norm
or the standard deviation (across features, for each example).
Parameters
----------
X : ndarray, 2-dimensional
Design matrix with examples indexed on the first axis and \
features indexed on the second.
scale : float, optional
Multiply features by this const.
subtract_mean : bool, optional
Remove the mean across features/pixels before normalizing. \
Defaults to `True`.
use_std : bool, optional
Normalize by the per-example standard deviation across features \
instead of the vector norm. Defaults to `False`.
sqrt_bias : float, optional
Fudge factor added inside the square root. Defaults to 0.
min_divisor : float, optional
If the divisor for an example is less than this value, \
do not apply it. Defaults to `1e-8`.
Returns
-------
Xp : ndarray, 2-dimensional
The contrast-normalized features.
Notes
-----
`sqrt_bias` = 10 and `use_std = True` (and defaults for all other
parameters) corresponds to the preprocessing used in [1].
References
----------
.. [1] A. Coates, H. Lee and A. Ng. "An Analysis of Single-Layer
Networks in Unsupervised Feature Learning". AISTATS 14, 2011.
http://www.stanford.edu/~acoates/papers/coatesleeng_aistats_2011.pdf
"""
assert X.ndim == 2, "X.ndim must be 2"
scale = float(scale)
assert scale >= min_divisor
# Note: this is per-example mean across pixels, not the
# per-pixel mean across examples. So it is perfectly fine
# to subtract this without worrying about whether the current
# object is the train, valid, or test set.
mean = X.mean(axis=1)
if subtract_mean:
X = X - mean[:, numpy.newaxis] # Makes a copy.
else:
X = X.copy()
if use_std:
# ddof=1 simulates MATLAB's var() behaviour, which is what Adam
# Coates' code does.
ddof = 1
# If we don't do this, X.var will return nan.
if X.shape[1] == 1:
ddof = 0
normalizers = numpy.sqrt(sqrt_bias + X.var(axis=1, ddof=ddof)) / scale
else:
normalizers = numpy.sqrt(sqrt_bias + (X ** 2).sum(axis=1)) / scale
# Don't normalize by anything too small.
normalizers[normalizers < min_divisor] = 1.
X /= normalizers[:, numpy.newaxis] # Does not make a copy.
return X
def applyGCN(save_image=False):
data = np.loadtxt("../inputData/train.csv", dtype=np.float32, delimiter=',', skiprows=1)
test_data = np.loadtxt("../inputData/test.csv", dtype=np.float32, delimiter=',', skiprows=1)
trainingFolder = "../inputData/converted_training/GCN/"
testingFolder = "../inputData/converted_testing/GCN/"
aux_data = data.copy()
###################################################
#############TRAIN_DATA############################
###################################################
# GCN#
data_to_gcn = data[:, 1:]
img_gcn = global_contrast_normalize(data_to_gcn)
for x in xrange(0, len(data[:, 1])):
print "Saving training image :", x
image_aux = np.copy(img_gcn[x, :])
image_aux = (image_aux).astype('uint8') * 255
aux_data[x, 1:] = image_aux.tolist()
if save_image is True:
mpimg.imsave(trainingFolder + "GCN_Gray_TrainImage_" + str(x), image_aux, cmap=plt.get_cmap('gray'))
# figsize = [x / float(dpi) for x in (arr.shape[1], arr.shape[0])]
# IndexError: tuple index out of range
with open(trainingFolder + "GCN_traindata.csv", 'wb') as fp:
for i in range(0, aux_data.shape[0]):
column = aux_data[i, :].tolist()
column = map(lambda x: str(x) + ',', column)
column = ''.join(column)[0:-1]
fp.write(column + '\n')
##################################################
#############TEST_DATA############################
##################################################
img_gcn = global_contrast_normalize(test_data[:, :])
aux_data = test_data.copy()
for x in xrange(0, len(test_data[:, 0])):
print "Saving testing image :", x
image_aux = np.copy(img_gcn[x, :])
image_aux = image_aux.astype('uint8') * 255
aux_data[x, :] = image_aux.tolist()
if save_image is True:
mpimg.imsave(testingFolder + "GCN_Gray_TestImage_" + str(x), image_aux, cmap=plt.get_cmap('gray'))
with open(testingFolder + "GCN_testdata.csv", 'wb') as fp:
for i in range(0, aux_data.shape[0]):
column = aux_data[i, :].tolist()
column = map(lambda x: str(x) + ',', column)
column = ''.join(column)[0:-1]
fp.write(column + '\n')
问题似乎出在这一行：
if save_image is True:
mpimg.imsave(trainingFolder + "GCN_Gray_TrainImage_" + str(x), image_aux, cmap=plt.get_cmap('gray'))
这只是将图像映射到灰度级后保存图像。
但我看不出错误与这条线有何关系......
我需要一些指导或提示。
欢迎任何帮助！
用户帮我修复后修复。
for x in xrange(0, len(test_data[:, 0])):
print "Saving testing image :", x
image_aux = np.copy(img_gcn[x, :])
aux_data[x, :] = image_aux.tolist()
image_aux = np.reshape(image_aux, (28,28))
if save_image is True:
mpimg.imsave(testingFolder + "GCN_Gray_TestImage_" + str(x), image_aux, cmap=plt.get_cmap('gray'))
将代码更改为此后已修复：
for x in xrange(0, len(test_data[:, 0])):
print "Saving testing image :", x
image_aux = np.copy(img_gcn[x, :])
aux_data[x, :] = image_aux.tolist()
image_aux = np.reshape(image_aux, (28,28))
if save_image is True:
mpimg.imsave(testingFolder + "GCN_Gray_TestImage_" + str(x), image_aux, cmap=plt.get_cmap('gray'))
我正在尝试将 GCN 应用于图像列表，但我一直收到此异常，
Saving training image : 0
Traceback (most recent call last):
File "A:/Code/MeusProjetosPython/KaggleMnistProject/Pre_Processing/Main.py", line 12, in <module>
applyGCN(True)
File "A:\Code\MeusProjetosPython\KaggleMnistProject\Pre_Processing\Global_Contrast_Normalization.py", line 123, in applyGCN
mpimg.imsave(trainingFolder + "GCN_Gray_TrainImage_" + str(x), image_aux, cmap=plt.get_cmap('gray'))
File "C:\SciSoft\WinPython-64bit-2.7.9.4\python-2.7.9.amd64\lib\site-packages\matplotlib\image.py", line 1310, in imsave
figsize = [x / float(dpi) for x in (arr.shape[1], arr.shape[0])]
IndexError: tuple index out of range
由于注释，完整代码有点大。但它只是两种方法。
不管怎样，我把代码粘贴在这里。
import numpy
import matplotlib.pyplot as plt
import matplotlib.image as mpimg
import numpy as np
def global_contrast_normalize(X, scale=1., subtract_mean=True, use_std=False, sqrt_bias=0., min_divisor=1e-8):
"""
Global contrast normalizes by (optionally) subtracting the mean
across features and then normalizes by either the vector norm
or the standard deviation (across features, for each example).
Parameters
----------
X : ndarray, 2-dimensional
Design matrix with examples indexed on the first axis and \
features indexed on the second.
scale : float, optional
Multiply features by this const.
subtract_mean : bool, optional
Remove the mean across features/pixels before normalizing. \
Defaults to `True`.
use_std : bool, optional
Normalize by the per-example standard deviation across features \
instead of the vector norm. Defaults to `False`.
sqrt_bias : float, optional
Fudge factor added inside the square root. Defaults to 0.
min_divisor : float, optional
If the divisor for an example is less than this value, \
do not apply it. Defaults to `1e-8`.
Returns
-------
Xp : ndarray, 2-dimensional
The contrast-normalized features.
Notes
-----
`sqrt_bias` = 10 and `use_std = True` (and defaults for all other
parameters) corresponds to the preprocessing used in [1].
References
----------
.. [1] A. Coates, H. Lee and A. Ng. "An Analysis of Single-Layer
Networks in Unsupervised Feature Learning". AISTATS 14, 2011.
http://www.stanford.edu/~acoates/papers/coatesleeng_aistats_2011.pdf
"""
assert X.ndim == 2, "X.ndim must be 2"
scale = float(scale)
assert scale >= min_divisor
# Note: this is per-example mean across pixels, not the
# per-pixel mean across examples. So it is perfectly fine
# to subtract this without worrying about whether the current
# object is the train, valid, or test set.
mean = X.mean(axis=1)
if subtract_mean:
X = X - mean[:, numpy.newaxis] # Makes a copy.
else:
X = X.copy()
if use_std:
# ddof=1 simulates MATLAB's var() behaviour, which is what Adam
# Coates' code does.
ddof = 1
# If we don't do this, X.var will return nan.
if X.shape[1] == 1:
ddof = 0
normalizers = numpy.sqrt(sqrt_bias + X.var(axis=1, ddof=ddof)) / scale
else:
normalizers = numpy.sqrt(sqrt_bias + (X ** 2).sum(axis=1)) / scale
# Don't normalize by anything too small.
normalizers[normalizers < min_divisor] = 1.
X /= normalizers[:, numpy.newaxis] # Does not make a copy.
return X
def applyGCN(save_image=False):
data = np.loadtxt("../inputData/train.csv", dtype=np.float32, delimiter=',', skiprows=1)
test_data = np.loadtxt("../inputData/test.csv", dtype=np.float32, delimiter=',', skiprows=1)
trainingFolder = "../inputData/converted_training/GCN/"
testingFolder = "../inputData/converted_testing/GCN/"
aux_data = data.copy()
###################################################
#############TRAIN_DATA############################
###################################################
# GCN#
data_to_gcn = data[:, 1:]
img_gcn = global_contrast_normalize(data_to_gcn)
for x in xrange(0, len(data[:, 1])):
print "Saving training image :", x
image_aux = np.copy(img_gcn[x, :])
image_aux = (image_aux).astype('uint8') * 255
aux_data[x, 1:] = image_aux.tolist()
if save_image is True:
mpimg.imsave(trainingFolder + "GCN_Gray_TrainImage_" + str(x), image_aux, cmap=plt.get_cmap('gray'))
# figsize = [x / float(dpi) for x in (arr.shape[1], arr.shape[0])]
# IndexError: tuple index out of range
with open(trainingFolder + "GCN_traindata.csv", 'wb') as fp:
for i in range(0, aux_data.shape[0]):
column = aux_data[i, :].tolist()
column = map(lambda x: str(x) + ',', column)
column = ''.join(column)[0:-1]
fp.write(column + '\n')
##################################################
#############TEST_DATA############################
##################################################
img_gcn = global_contrast_normalize(test_data[:, :])
aux_data = test_data.copy()
for x in xrange(0, len(test_data[:, 0])):
print "Saving testing image :", x
image_aux = np.copy(img_gcn[x, :])
image_aux = image_aux.astype('uint8') * 255
aux_data[x, :] = image_aux.tolist()
if save_image is True:
mpimg.imsave(testingFolder + "GCN_Gray_TestImage_" + str(x), image_aux, cmap=plt.get_cmap('gray'))
with open(testingFolder + "GCN_testdata.csv", 'wb') as fp:
for i in range(0, aux_data.shape[0]):
column = aux_data[i, :].tolist()
column = map(lambda x: str(x) + ',', column)
column = ''.join(column)[0:-1]
fp.write(column + '\n')
问题似乎出在这一行：
if save_image is True:
mpimg.imsave(trainingFolder + "GCN_Gray_TrainImage_" + str(x), image_aux, cmap=plt.get_cmap('gray'))
这只是将图像映射到灰度级后保存图像。 但我看不出错误与这条线有何关系...... 我需要一些指导或提示。 欢迎任何帮助！
用户帮我修复后修复。
for x in xrange(0, len(test_data[:, 0])):
print "Saving testing image :", x
image_aux = np.copy(img_gcn[x, :])
aux_data[x, :] = image_aux.tolist()
image_aux = np.reshape(image_aux, (28,28))
if save_image is True:
mpimg.imsave(testingFolder + "GCN_Gray_TestImage_" + str(x), image_aux, cmap=plt.get_cmap('gray'))
将代码更改为此后已修复：
for x in xrange(0, len(test_data[:, 0])):
print "Saving testing image :", x
image_aux = np.copy(img_gcn[x, :])
aux_data[x, :] = image_aux.tolist()
image_aux = np.reshape(image_aux, (28,28))
if save_image is True:
mpimg.imsave(testingFolder + "GCN_Gray_TestImage_" + str(x), image_aux, cmap=plt.get_cmap('gray'))