Keras VGG16 預訓練模型准確率沒有提高

Question

大家好，我在 Keras 上的 VGG16 上遇到了問題。
我試圖提高准確性，但沒有奏效。

我只有 46 個數據訓練、12 個類和 26 個數據驗證。

目前，我能得到的最高精度是 0.18。 我嘗試將批量大小更改為 2，但結果比我預期的要差。 我不認為我應該設置數據訓練樣本應該高於我的實際數據。

我應該怎么做才能提高准確性？

這是我的實際代碼：

from keras.applications.vgg16 import VGG16
from keras.preprocessing import image
from keras.applications.vgg16 import preprocess_input
from keras.layers import Input, Flatten, Dense, Dropout
from keras.models import Model, Sequential
from keras import backend as K
import numpy as np
import matplotlib.pyplot as plt

# dimensions of our images.
from keras.preprocessing.image import ImageDataGenerator

img_width, img_height = 224, 224

train_data_dir = 'database/train'
validation_data_dir = 'database/validation'
nb_train_samples = 46
nb_validation_samples = 26
epochs = 50
batch_size = 4

if K.image_data_format() == 'channels_first':
    input_shape = (3, img_width, img_height)
else:
    input_shape = (img_width, img_height, 3)


#Get back the convolutional part of a VGG network trained on ImageNet
vgg_conv = VGG16(weights='imagenet', include_top=True)
vgg_conv.summary()
print('VGG Pretrained Model loaded.')

#Add a layer where input is the output of the  second last layer
x = Dense(12, activation='softmax', name='predictions')(vgg_conv.layers[-2].output)

model = Model(input=vgg_conv.input, output=x)
#In the summary, weights and layers from VGG part will be hidden, but they will be fit during the training
model.summary()


# this is the augmentation configuration we will use for training
train_datagen = ImageDataGenerator(
    rescale=1. / 224,
    shear_range=0.2,
    zoom_range=0.2,
    horizontal_flip=True)

# this is the augmentation configuration we will use for testing:
# only rescaling
test_datagen = ImageDataGenerator(rescale=1. / 224)

train_generator = train_datagen.flow_from_directory(
    train_data_dir,
    target_size=(img_width, img_height),
    batch_size=batch_size,
    class_mode='categorical')

validation_generator = test_datagen.flow_from_directory(
    validation_data_dir,
    target_size=(img_width, img_height),
    batch_size=batch_size,
    class_mode='categorical')

# compile model
# model.compile(loss='sparse_categorical_crossentropy', optimizer=optimizers.RMSprop(lr=2e-4), metrics=['accuracy'])
model.compile(loss='categorical_crossentropy', optimizer='rmsprop', metrics=['accuracy'])

# Train the model
history = model.fit_generator(
    train_generator,
    steps_per_epoch=nb_train_samples / batch_size,
    epochs=epochs,
    validation_data=validation_generator,
    validation_steps=nb_validation_samples / batch_size)

# Save the model
model.save('vgg16_pretrained_5.h5')

# Check Performance
acc = history.history['acc']
val_acc = history.history['val_acc']
loss = history.history['loss']
val_loss = history.history['val_loss']

epochs = range(len(acc))

plt.plot(epochs, acc, 'b', label='Training acc')
plt.plot(epochs, val_acc, 'r', label='Validation acc')
plt.title('Training and validation accuracy')
plt.legend()

plt.figure()

plt.plot(epochs, loss, 'b', label='Training loss')
plt.plot(epochs, val_loss, 'r', label='Validation loss')
plt.title('Training and validation loss')
plt.legend()

plt.show()

Answer 1

由於您有 12 個類並且只有 46 個觀測值，因此每個類大致變為 2 個觀測值（這只是猜測，甚至沒有查看數據集）。 有了這么少的數據，NN 模型甚至無法理解數據的模式，最終將無法泛化。 因此，您至少需要超過 2k 次觀察才能獲得更好的結果。