Keras对Python的预测

Question

I have this code for my CNN:

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

# dimensions of our images.
img_width, img_height = 64, 64

train_data_dir = "path_trainning"
validation_data_dir = "path_validation"
nb_train_samples = 2000
nb_validation_samples = 800
epochs = 10
batch_size = 16

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

model = Sequential()
model.add(Conv2D(32, (3, 3), input_shape=input_shape))
model.add(Activation('relu'))
model.add(MaxPooling2D(pool_size=(2, 2)))

model.add(Conv2D(32, (3, 3)))
model.add(Activation('relu'))
model.add(MaxPooling2D(pool_size=(2, 2)))

model.add(Conv2D(64, (3, 3)))
model.add(Activation('relu'))
model.add(MaxPooling2D(pool_size=(2, 2)))

model.add(Flatten())
model.add(Dense(64))
model.add(Activation('relu'))
model.add(Dropout(0.5))
model.add(Dense(1))
model.add(Activation('sigmoid'))

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('my_cnn.h5')

And this is the code for my predictions:

for file in os.listdir(targets_path):
    filef = '\\' + file
    test_image = image.load_img(targets_path + filef, target_size=(64, 64))
    test_image = image.img_to_array(test_image)
    test_image = np.expand_dims(test_image, axis=0)
    result = model.predict(test_image)
    print("\nOriginal: " + file)
    print("Prediction: " + str(result[0][0]))
    if result[0][0] == 1:
        prediction = 'dog'
    else:
        prediction = 'cat'
    print(prediction)

My question is:

With this code as the "Prediction" part, I am realising that unless the CNN has a 1, it won't be a dog. And I am getting results like 0.99999 is a cat, but with that value it is closer to be a dog.

I think I am not understanding it properly.

Could someone explain me please?

Answer 1

这是由于输出层是具有S型激活的节点，该节点返回0到1之间的值。因此，结果永远不会为1（或0），因此代码将始终返回“ cat”。

Answer 2

This might be the issue with your CNN.

You are using an ReLU activation in the hidden layers. They have an output range from 0 to Infinity. When these values flow through the final output activation which is sigmoid in your case. If I pass a greater value like 25 to sigmoid the output will be close to 1. The same would happen with a very small value which would result in threshold closer to 0.

You should use a softmax function at the output layer if you are using ReLU in the hidden layers. Softmax converts the logits to class probabilities.

And also, with softmax, you would use categorical classes and not binary. You will have 2 classes and hence 2 output nodes.