使用 SVM 分类器作为预训练模型的最后一层（VGG16）答案

【问题标题】：Using SVM classifier as the last layer for a Pre-trained Model(VGG16)使用 SVM 分类器作为预训练模型的最后一层（VGG16）
【发布时间】：2021-10-14 16:23:51
【问题描述】：

我已经使用 Keras 训练了我的 CNN 模型（二元分类），现在我想使用 SVM 分类器而不是使用全连接层进行分类。

我使用 VGG16 预训练网络进行特征提取，还使用了数据增强。

添加SVM作为分类的最后一层的可能方法是什么？

#Parametres  
import keras
from keras.applications import VGG16
import sys
from PIL import Image

#Using VGG16 Pre-trained Model
conv_base = VGG16(weights = 'imagenet',
              include_top = False,
              input_shape=(224, 224, 3))

conv_base.summary()

import numpy as np
import os
from keras.preprocessing.image import ImageDataGenerator

base_dir = 'C:Covid Detection/Code/Dataset-created')
train_dir = os.path.join(base_dir, 'train')
validation_dir = os.path.join(base_dir, 'validation')
test_dir = os.path.join(base_dir, 'test')

from keras import models
from keras import layers

model = models.Sequential()
model.add(conv_base)
model.add(layers.Flatten())
model.add(layers.Dense(256, activation='relu'))
model.add(layers.Dense(1, activation='sigmoid'))

conv_base.trainable = False

from keras import optimizers

train_datagen = ImageDataGenerator(
    rescale=1./255,
    rotation_range=40,
    width_shift_range=0.2,
    height_shift_range=0.2,
    shear_range=0.2,
    zoom_range=0.2,
    horizontal_flip=True,
    fill_mode='nearest')

test_datagen = ImageDataGenerator(rescale=1./255)

train_generator = train_datagen.flow_from_directory(
    train_dir,
    target_size=(224, 224),
    batch_size=20,
    class_mode='binary')

validation_generator = test_datagen.flow_from_directory(
    validation_dir,
    target_size=(224, 224),
    batch_size=20,
    class_mode='binary')

# Compile the model
model.compile(loss='binary_crossentropy',
optimizer=optimizers.RMSprop(lr=2e-5),
metrics=['acc'])

# Train the model
history = model.fit_generator(
    train_generator,
    steps_per_epoch=50,
    epochs=30,
    validation_data=validation_generator,
    validation_steps=50)

# Save the model
model.save('vgg16_aug.h5')

【问题讨论】：

标签： keras

【解决方案1】：

from keras.applications.vgg16 import VGG16
from keras.preprocessing.image import load_img
from keras.preprocessing.image import img_to_array
from keras.applications.vgg16 import preprocess_input


base_model = VGG16(weights='imagenet')
model = Model(inputs=base_model.input,      
outputs=base_model.get_layer('flatten').output)
model.summary()

创建一个使用 VGG16 提取特征的函数

def get_features(img_path):
    img = load_img(img_path, target_size=(224, 224))
    x = img_to_array(img)
    x = np.expand_dims(x, axis=0)
    x = preprocess_input(x)
    flatten = model.predict(x)
    return list(flatten[0])

features, labels = [], []

# Loop into the directory of images and extract features and labels
for image_path in folder:
    features.append(extract_features(image_path))
    labels.append("#some label")

直到这里我们提取了特征和标签，现在我们使用 SVM 训练这些特征如下：-

from sklearn.model_selection import train_test_split
from sklearn.svm import LinearSVC
from sklearn.metrics import accuracy_score

X_train, X_test, y_train, y_test = train_test_split( features, 
                                                     labels, 
                                                     test_size=0.30)
                                         

clf = LinearSVC(random_state=0, tol=1e-5)
clf.fit(X_train, y_train)

predicted = clf.predict(X_test)

# get the accuracy
print(accuracy_score(y_test, predicted))

【讨论】：