Question

我正在尝试解决21类分类问题。这是代码：

# dimensions of our images.
img_width, img_height = 256, 256

top_model_weights_path = 'bottleneck_fc_model1.h5'
train_data_dir = 'data1/train1'
validation_data_dir = 'data1/validation1'
nb_train_samples = 1680
nb_validation_samples = 420
epochs = 10
batch_size = 16


def save_bottlebeck_features():
    datagen = ImageDataGenerator(rescale=1. / 255)

    # build the VGG16 network
    model = applications.VGG16(include_top=False, weights='imagenet')

    generator = datagen.flow_from_directory(
        train_data_dir,
        target_size=(img_width, img_height),
        batch_size=batch_size,
        class_mode=None,
        shuffle=False)
    bottleneck_features_train = model.predict_generator(
        generator, nb_train_samples // batch_size)
    np.save('bottleneck_features_train1.npy',
            bottleneck_features_train)

    generator = datagen.flow_from_directory(
        validation_data_dir,
        target_size=(img_width, img_height),
        batch_size=batch_size,
        class_mode=None,
        shuffle=False)
    bottleneck_features_validation = model.predict_generator(
        generator, nb_validation_samples // batch_size)
    np.save('bottleneck_features_validation1.npy',
            bottleneck_features_validation)


def train_top_model():
    train_data = np.load('bottleneck_features_train1.npy')
    train_labels = np.zeros((1680,21))
    j = 0
    i = 0
    for j in range(0, 21):
        train_labels[i:i+80, j] = 1
        i = i+80

    validation_data = np.load('bottleneck_features_validation1.npy')
    validation_labels = np.zeros((420,21))
    j = 0
    i = 0
    for j in range(0, 21):
        validation_labels[i:i+20, j] = 1
        i = i+20

    model = Sequential()
    model.add(Flatten(input_shape=train_data.shape[1:]))
    model.add(Dense(256, activation='relu'))
    model.add(Dropout(0.5))
    model.add(Dense(21, activation='softmax'))

    model.compile(optimizer='rmsprop',
                  loss='categorical_crossentropy', metrics=['accuracy'])

    model.fit(train_data, train_labels,
              epochs=epochs,
              batch_size=batch_size,
              validation_data=(validation_data, validation_labels))
    model.save_weights(top_model_weights_path)


save_bottlebeck_features()
train_top_model()

我不断收到一条错误消息，说我没有与验证目标样本相同数量的验证输入样本：

  File "<ipython-input-96-0da2181ac8b3>", line 1, in <module>
    train_top_model()

  File "<ipython-input-87-43a97663639c>", line 36, in train_top_model
    validation_data=(validation_data, validation_labels))

  File "C:\Users\Spencer\Anaconda3_2\envs\tensorflow\lib\site-packages\keras\engine\training.py", line 972, in fit
    batch_size=batch_size)

  File "C:\Users\Spencer\Anaconda3_2\envs\tensorflow\lib\site-packages\keras\engine\training.py", line 804, in _standardize_user_data
    check_array_length_consistency(x, y, sample_weights)

  File "C:\Users\Spencer\Anaconda3_2\envs\tensorflow\lib\site-packages\keras\engine\training_utils.py", line 237, in check_array_length_consistency
    'and ' + str(list(set_y)[0]) + ' target samples.')

ValueError: Input arrays should have the same number of samples as target arrays. Found 416 input samples and 420 target samples.

我不知道为什么要说我有416个输入样本和420个目标样本；我肯定有420个输入样本和420个目标样本。关于模型为何如此表现的任何想法？

Answer 1

问题是@RequestMapping(method = RequestMethod.DELETE) public ResponseEntity<Void> methodName() { // your buisness logic return new ResponseEntity.noContent().build(); }（16）无法将nb_validation_samples（420）整除。这导致我的batch_size调用按以下方式执行楼层划分：420 // 16 ==26。因此，样本总数等于16 * 26 == 416。

我将predict_generator()更改为10，现在一切运行正常。

ValueError：输入数组应具有与目标数组相同数量的样本。找到416个输入样本和420个目标样本

1 个答案: