Classification test in Scikit-learn, ValueError: setting an array element with a sequence

Question

Using the tutorial on multiclass adaboost, I'm trying to classify some images that have two classes (but I don't suppose the algorithm shouldn't work if the problem is binary). Then I'm going to extend my samples to include other classes.

My current test is quite small, only 17 images in all, 10 for training, 7 for testing.

For now I have two classes: 0: no vehicle, 1: vehicle present I used integer labels because according to the example in the link above, the training data consists of integer-based labels.

I've edited the provided example only a bit, to include my own image files, but I'm getting an error.

Traceback (most recent call last):
  File "C:\Users\app\Documents\Python Scripts\carclassify.py", line 66, in <module>
    bdt_discrete.fit(X_train, y_train)
  File "C:\Users\app\Anaconda\lib\site-packages\sklearn\ensemble\weight_boosting.py", line 389, in fit
    return super(AdaBoostClassifier, self).fit(X, y, sample_weight)
  File "C:\Users\app\Anaconda\lib\site-packages\sklearn\ensemble\weight_boosting.py", line 99, in fit
    X = np.ascontiguousarray(array2d(X), dtype=DTYPE)
  File "C:\Users\app\Anaconda\lib\site-packages\numpy\core\numeric.py", line 408, in ascontiguousarray
    return array(a, dtype, copy=False, order='C', ndmin=1)
ValueError: setting an array element with a sequence.

The following is my code, adapted from the example on the scikit-learn website:

f = open("PATH_TO_SAMPLES\\samples.txt",'r')
out = f.read().splitlines()
import numpy as np

imgs = []
tmp_hogs = []
# 13 of the images are with vehicles, 4 are without
labels = [1,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0]

for file in out:
        filepath = "C:\PATH_TO_SAMPLE_IMAGES\\" + file
        curr_img = color.rgb2gray(io.imread(filepath))
        imgs.append(resize(curr_img,(60,40)))
        fd, hog_image = hog(curr_img, orientations=8, pixels_per_cell=(16, 16),
                 cells_per_block=(1, 1), visualise=True)
        tmp_hogs.append(fd) 

img_hogs = np.array(tmp_hogs)
n_split = 10
X_train, X_test = img_hogs[:n_split], X[n_split:] # all first ten images with vehicles
y_train, y_test = labels[:n_split], labels[n_split:] # 3 images with vehicles, 4 without

#now all the code below is straight off the example on scikit-learn's website

bdt_real = AdaBoostClassifier(
    DecisionTreeClassifier(max_depth=2),
    n_estimators=600,
    learning_rate=1)

bdt_discrete = AdaBoostClassifier(
    DecisionTreeClassifier(max_depth=2),
    n_estimators=600,
    learning_rate=1.5,
    algorithm="SAMME")

bdt_real.fit(X_train, y_train)
bdt_discrete.fit(X_train, y_train)

real_test_errors = []
discrete_test_errors = []

for real_test_predict, discrete_train_predict in zip(
        bdt_real.staged_predict(X_test), bdt_discrete.staged_predict(X_test)):
    real_test_errors.append(
        1. - accuracy_score(real_test_predict, y_test))
    discrete_test_errors.append(
        1. - accuracy_score(discrete_train_predict, y_test))

n_trees = xrange(1, len(bdt_discrete) + 1)

pl.figure(figsize=(15, 5))

pl.subplot(131)
pl.plot(n_trees, discrete_test_errors, c='black', label='SAMME')
pl.plot(n_trees, real_test_errors, c='black',
        linestyle='dashed', label='SAMME.R')
pl.legend()
pl.ylim(0.18, 0.62)
pl.ylabel('Test Error')
pl.xlabel('Number of Trees')

pl.subplot(132)
pl.plot(n_trees, bdt_discrete.estimator_errors_, "b", label='SAMME', alpha=.5)
pl.plot(n_trees, bdt_real.estimator_errors_, "r", label='SAMME.R', alpha=.5)
pl.legend()
pl.ylabel('Error')
pl.xlabel('Number of Trees')
pl.ylim((.2,
        max(bdt_real.estimator_errors_.max(),
            bdt_discrete.estimator_errors_.max()) * 1.2))
pl.xlim((-20, len(bdt_discrete) + 20))

pl.subplot(133)
pl.plot(n_trees, bdt_discrete.estimator_weights_, "b", label='SAMME')
pl.legend()
pl.ylabel('Weight')
pl.xlabel('Number of Trees')
pl.ylim((0, bdt_discrete.estimator_weights_.max() * 1.2))
pl.xlim((-20, len(bdt_discrete) + 20))

# prevent overlapping y-axis labels
pl.subplots_adjust(wspace=0.25)
pl.show()

Edit

I typed

print tmp_hogs

and the output was this:

[ array([ 0.27621208,  0.11038658,  0.10698133, ...,  0.08661556,        0.04612063,  0.0280782 ]), 
        array([  0.00000000e+00,   0.00000000e+00,   0.00000000e+00, ..., -1.29909838e-15,  -7.01780982e-17,  -1.24900943e-15]), 
        array([ 0.0503603 ,  0.1497235 ,  0.2372957 , ...,  0.07249325, 0.04545541,  0.00903818]), 
        array([ 0.27299191,  0.13122109,  0.0719268 , ...,  0.0848522 ,  0.04789403,  0.01387038]), 
        array([  0.00000000e+00,   0.00000000e+00,   0.00000000e+00, ...,  3.32140617e-17,  -6.58924128e-17,  -6.23567224e-16]), 
        array([ 0.37431874,  0.18094303,  0.01219871, ...,  0.06501856, 0.04855516,  0.02439321]), 
        array([ 0.41087302,  0.16478851,  0.03396399, ...,  0.09511273, 0.04077713,  0.03945513]), 
        array([ 0.17753915,  0.07025565,  0.09136909, ...,  0.03396507, 0.01379266,  0.01645722]), 
        array([ 0.40605587,  0.05915388,  0.03767763, ...,  0.08981079, 0.05452031,  0.01725399]), 
        array([ 0.        ,  0.        ,  0.        , ...,  0.00579303, 0.02053979,  0.0019091 ]), 
        array([ 0.31550735,  0.11988131,  0.07716529, ...,  0.09815158, 0.03058497,  0.02236517]), 
        array([  0.00000000e+00,   0.00000000e+00,   0.00000000e+00, ..., -3.51175682e-16,   1.31619418e-03,   2.86127901e-16]), 
        array([ 0.21381704,  0.22352378,  0.11568828, ...,  0.06311083, 0.02696666,  0.00402261]), 
        array([ 0.17480064,  0.1469145 ,  0.16336016, ...,  0.05614001, 0.03244093,  0.00524034]), 
        array([ 0.        ,  0.        ,  0.        , ...,  0.03089959, 0.00509584,  0.00247698]), 
        array([ 0.04711166,  0.0218663 ,  0.05316   , ...,  0.04214594, 0.04892439,  0.25840958]), 
        array([ 0.05357464,  0.00530857,  0.07162301, ...,  0.06802692, 0.08331959,  0.26619977])]

Then I ran

print img_hogs

and the output was:

[ array([ 0.27621208,  0.11038658,  0.10698133, ...,  0.08661556, 0.04612063,  0.0280782 ])
 array([  0.00000000e+00,   0.00000000e+00,   0.00000000e+00, ..., -1.29909838e-15,  -7.01780982e-17,  -1.24900943e-15])
 array([ 0.0503603 ,  0.1497235 ,  0.2372957 , ...,  0.07249325, 0.04545541,  0.00903818])
 array([ 0.27299191,  0.13122109,  0.0719268 , ...,  0.0848522 , 0.04789403,  0.01387038])
 array([  0.00000000e+00,   0.00000000e+00,   0.00000000e+00, ..., 3.32140617e-17,  -6.58924128e-17,  -6.23567224e-16])
 array([ 0.37431874,  0.18094303,  0.01219871, ...,  0.06501856, 0.04855516,  0.02439321])
 array([ 0.41087302,  0.16478851,  0.03396399, ...,  0.09511273, 0.04077713,  0.03945513])
 array([ 0.17753915,  0.07025565,  0.09136909, ...,  0.03396507, 0.01379266,  0.01645722])
 array([ 0.40605587,  0.05915388,  0.03767763, ...,  0.08981079, 0.05452031,  0.01725399])
 array([ 0.        ,  0.        ,  0.        , ...,  0.00579303, 0.02053979,  0.0019091 ])
 array([ 0.31550735,  0.11988131,  0.07716529, ...,  0.09815158, 0.03058497,  0.02236517])
 array([  0.00000000e+00,   0.00000000e+00,   0.00000000e+00, ..., -3.51175682e-16,   1.31619418e-03,   2.86127901e-16])
 array([ 0.21381704,  0.22352378,  0.11568828, ...,  0.06311083, 0.02696666,  0.00402261])
 array([ 0.17480064,  0.1469145 ,  0.16336016, ...,  0.05614001, 0.03244093,  0.00524034])
 array([ 0.        ,  0.        ,  0.        , ...,  0.03089959, 0.00509584,  0.00247698])
 array([ 0.04711166,  0.0218663 ,  0.05316   , ...,  0.04214594, 0.04892439,  0.25840958])
 array([ 0.05357464,  0.00530857,  0.07162301, ...,  0.06802692, 0.08331959,  0.26619977])]

Answer 1

try:

imgs = []
tmp_hogs = np.zeros((17, 256))
# 13 of the images are with vehicles, 4 are without
labels = [1,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0]

i = 0
for file in out:
        filepath = "C:\PATH_TO_SAMPLE_IMAGES\\" + file
        curr_img = color.rgb2gray(io.imread(filepath))
        imgs.append(resize(curr_img,(60,40)))
        fd, hog_image = hog(curr_img, orientations=8, pixels_per_cell=(16, 16),
                 cells_per_block=(1, 1), visualise=True)
        tmp_hogs[i,:] = fd
        i+=1

img_hogs = tmp_hogs