图像分类器的混淆矩阵和F1分数

Question

hiya我遵循本指南，了解如何使用少量数据制作功能强大的图像分类器 https://blog.keras.io/building-powerful-image-classification-models-using-very-little-data.html 我需要制作一个融合矩阵并获得此代码的F1分数，这是一个图像分类器，可以检测出所有数据集都是灰度的肿瘤

from keras.preprocessing.image import ImageDataGenerator
from keras.models import Sequential
from keras.layers import Conv2D, MaxPooling2D
from keras.layers import Activation, Dropout, Flatten, Dense
from keras import backend as K
from PIL import ImageFile, Image
print(Image.__file__)
import numpy as np
import matplotlib.pyplot as plt

# dimensions of our images.
img_width, img_height = 150, 150

train_data_dir = r'C:\Users\Acer\imagerec\Brain\TRAIN'
validation_data_dir = r'C:\Users\Acer\imagerec\Brain\VAL'
nb_train_samples = 140
nb_validation_samples = 40
epochs = 20
batch_size = 5

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

from keras.applications.vgg16 import VGG16
from keras.models import Model
from keras.layers import Dense

vgg = VGG16(include_top=False, weights='imagenet', input_shape=(), pooling='avg')
x = vgg.output
x = Dense(1, activation='sigmoid')(x)
model = Model(vgg.input, x)
model.summary()

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

# this is the augmentation configuration we will use for training
train_datagen = ImageDataGenerator(
    rescale=1. / 255,
    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. / 255)

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

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

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)

model.save_weights('first_try.h5')

编辑 现在我得到了这些代码

from keras.preprocessing.image import ImageDataGenerator
from keras.models import Sequential
from keras.layers import Conv2D, MaxPooling2D
from keras.layers import Activation, Dropout, Flatten, Dense
from keras import backend as K
from PIL import ImageFile, Image
print(Image.__file__)
import numpy
import matplotlib.pyplot as plt

# dimensions of our images.
img_width, img_height = 150, 150

train_data_dir = r'C:\Users\Acer\imagerec\Brain\TRAIN'
validation_data_dir = r'C:\Users\Acer\imagerec\Brain\VAL'
nb_train_samples = 140
nb_validation_samples = 40
epochs = 2
batch_size = 5

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

from keras.applications.vgg16 import VGG16
from keras.models import Model
from keras.layers import Dense

vgg = VGG16(include_top=False, weights='imagenet', input_shape=(), pooling='avg')
x = vgg.output
x = Dense(1, activation='sigmoid')(x)
model = Model(vgg.input, x)
model.summary()

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

# this is the augmentation configuration we will use for training
train_datagen = ImageDataGenerator(
    rescale=1. / 255,
    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. / 255)

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

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

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)

from sklearn.metrics import classification_report
from sklearn.metrics import confusion_matrix

test_steps_per_epoch = numpy.math.ceil(test_datagen.samples / test_datagen.batch_size)

predictions = model.predict_generator(validation_generator, steps=test_steps_per_epoch)
# Get most likely class
predicted_classes = numpy.argmax(predictions, axis=1)
true_classes = validation_generator.classes
class_labels = list(validation_generator.class_indices.keys())
report = classification_report(true_classes, predicted_classes, target_names=class_labels)
print(report)

cm=confusion_matrix(true_classes,predicted_classes)
print(cm)

model.save_weights('first_try.h5')

解决

from keras.preprocessing.image import ImageDataGenerator
from keras.models import Sequential
from keras.layers import Conv2D, MaxPooling2D
from keras.layers import Activation, Dropout, Flatten, Dense
from keras import backend as K
from PIL import ImageFile, Image
print(Image.__file__)
import numpy
import matplotlib.pyplot as plt

# dimensions of our images.
img_width, img_height = 150, 150

train_data_dir = r'C:\Users\Acer\imagerec\Brain\TRAIN'
validation_data_dir = r'C:\Users\Acer\imagerec\Brain\VAL'
nb_train_samples = 140
nb_validation_samples = 40
epochs = 2
batch_size = 5

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

from keras.applications.vgg16 import VGG16
from keras.models import Model
from keras.layers import Dense

vgg = VGG16(include_top=False, weights='imagenet', input_shape=(), pooling='avg')
x = vgg.output
x = Dense(1, activation='sigmoid')(x)
model = Model(vgg.input, x)
model.summary()

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

# this is the augmentation configuration we will use for training
train_datagen = ImageDataGenerator(
    rescale=1. / 255,
    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. / 255)

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

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

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)

from sklearn.metrics import classification_report
from sklearn.metrics import confusion_matrix

test_steps_per_epoch = numpy.math.ceil(validation_generator.samples / validation_generator.batch_size)

predictions = model.predict_generator(validation_generator, steps=test_steps_per_epoch)
# Get most likely class
predicted_classes = numpy.argmax(predictions, axis=1)
true_classes = validation_generator.classes
class_labels = list(validation_generator.class_indices.keys())
report = classification_report(true_classes, predicted_classes, target_names=class_labels)
print(report)

cm=confusion_matrix(true_classes,predicted_classes)
print(cm)

plt.imshow(cm)


model.save_weights('first_try.h5')

Answer 1

以下代码将为您的验证生成器生成混淆矩阵和分类报告

from sklearn.metrics import classification_report
from sklearn.metrics import confusion_matrix

predictions = model.predict_generator(validation_generator, steps=test_steps_per_epoch)
test_steps_per_epoch = numpy.math.ceil(validation_generator.samples / validation_generator.batch_size)
predicted_classes = numpy.argmax(predictions, axis=1)
true_classes = validation_generator.classes
class_labels = list(validation_generator.class_indices.keys())
report = classification_report(true_classes, predicted_classes, target_names=class_labels)
print(report)   

cm=confusion_matrix(true_classes,predicted_classes)
print(cm)

绘制使用情况

import matplotlib.pyplot as plt
plt.imshow(cm)