Question

我正在使用autokeras运行代码以找到最适合我的单元格数据的模型。这是针对运行autokeras 0.3.5的Linux Ubuntu系统16.04。输入数据的形状有问题。这是我的代码：

#define data directories
train_data_dir = 'malaria100/train' 
test_data_dir = 'malaria100/test'

# declare the number of samples in each category
nb_train_samples = 24760 
nb_test_samples = 2730 
num_classes = 2 
img_rows_orig = 100 
img_cols_orig = 100

def load_training_data():
    labels = os.listdir(train_data_dir)
    total = len(labels)
    X_train = np.ndarray((nb_train_samples, img_rows_orig, 
img_cols_orig, 3), dtype=np.uint8)
    Y_train = np.zeros((nb_train_samples,), dtype='uint8')
    i = 0
    print('-'*30)
    print('Creating training images...')
    print('-'*30)
    j = 0
    for label in labels:
        image_names_train = os.listdir(os.path.join(train_data_dir, 
label))
        total = len(image_names_train)
        print(label, total)
        for image_name in image_names_train:
            img = cv2.imread(os.path.join(train_data_dir, label, 
image_name), cv2.IMREAD_COLOR)
            img = np.array([img])
            X_train[i] = img
            Y_train[i] = j
            if i % 100 == 0:
                print('Done: {0}/{1} images'.format(i, total))
            i += 1
        j += 1    
    print(i)                
    print('Loading done.')
    print('Transform targets to keras compatible format.')
    Y_train = np_utils.to_categorical(Y_train[:nb_train_samples], 
num_classes)
    np.save('imgs_train.npy', X_train, Y_train) #save as numpy files
    return X_train, Y_train

def load_test_data():
    labels = os.listdir(test_data_dir)
    X_test = np.ndarray((nb_test_samples, img_rows_orig, img_cols_orig, 3), dtype=np.uint8)
    Y_test = np.zeros((nb_test_samples,), dtype='uint8')
    i = 0
    print('-'*30)
    print('Creating test images...')
    print('-'*30)
    j = 0
    for label in labels:
        image_names_test = os.listdir(os.path.join(test_data_dir, 
label))
        total = len(image_names_test)
        print(label, total)
        for image_name in image_names_test:
            img = cv2.imread(os.path.join(test_data_dir, label, 
image_name), cv2.IMREAD_COLOR)
            img = np.array([img])
            X_test[i] = img
            Y_test[i] = j
            if i % 100 == 0:
                print('Done: {0}/{1} images'.format(i, total))
            i += 1
         j += 1
    print(i)            
    print('Loading done.')
    print('Transform targets to keras compatible format.');
    Y_test = np_utils.to_categorical(Y_test[:nb_test_samples], 
num_classes)
    np.save('imgs_test.npy', X_test, Y_test) #save as numpy files
    return X_test, Y_test

def run():
    OUTPUT_PATH = "output"
    TRAINING_TIMES = [60 * 60, 60 * 60 * 2, 60 * 60 * 4, 60 * 60 * 
8,60 * 60 * 12, 60 * 60 * 24,]
    X_train, Y_train = load_training_data()
    X_test, Y_test = load_test_data()
    X_train = X_train.astype("float") / 255.0
    X_test = X_test.astype("float") / 255.0
    labelNames = ["abnormal", "normal"]

    for seconds in TRAINING_TIMES:
        print("[INFO] training model for {} seconds 
 max...".format(seconds))
        clf = ak.ImageClassifier(verbose=True, augment=False)
        clf.fit(X_train, Y_train, time_limit=seconds)
        clf.final_fit(X_train, Y_train, X_test, Y_test, 
retrain=True)
        # evaluate the Auto-Keras model
        score = clf.evaluate(X_test, Y_test)
        predictions = clf.predict(X_test)
        report = classification_report(Y_test, 
predictions,target_names=labelNames)
        p = os.path.sep.join(OUTPUT_PATH, "{}.txt".format(seconds))
        f = open(p, "w")
        f.write(report)
        f.write("\nscore: {}".format(score))
        f.close()

if __name__ == '__main__':
    run()

完整的错误列表如下所示：

  File "<ipython-input-4-f6faa80ae27f>", line 27, in <module>
run()

  File "<ipython-input-4-f6faa80ae27f>", line 13, in run
clf.fit(X_train, Y_train, time_limit=seconds)

  File "/home/rajaramans2/anaconda3/lib/python3.6/site-packages/autokeras/image/image_supervised.py", line 114, in fit
super().fit(x, y, time_limit)

  File "/home/rajaramans2/anaconda3/lib/python3.6/site-packages/autokeras/supervised.py", line 115, in fit
random_state=42)

  File "/home/rajaramans2/anaconda3/lib/python3.6/site-packages/sklearn/model_selection/_split.py", line 2184, in train_test_split
arrays = indexable(*arrays)

  File "/home/rajaramans2/anaconda3/lib/python3.6/site-packages/sklearn/utils/validation.py", line 260, in indexable
check_consistent_length(*result)

  File "/home/rajaramans2/anaconda3/lib/python3.6/site-packages/sklearn/utils/validation.py", line 235, in check_consistent_length
" samples: %r" % [int(l) for l in lengths])

ValueError: Found input variables with inconsistent numbers of samples: [24760, 49520]

输入数据的形状如下：

print(X_train.shape, X_test.shape, Y_train.shape, Y_test.shape)
(24760, 100, 100, 3) (2730, 100, 100, 3) (24760, 2) (2730, 2)

Answer 1

Autokeras图像分类器不接受分类标签。对此进行评论并保持Y的形状为（num_samples，1）即可解决问题。

ValueError：找到样本数量不一致的输入变量：[24760，49520]

1 个答案: