Question

我想问一个问题，关于我几天前就摔断了头的模型... 我正在尝试编写CNN模型来预测2类之间的结果（类似于猫或狗的问题）。我的代码可以识别“训练”和“验证”文件夹中的图片，也可以识别“测试”文件夹中要预测的图片，但是所有预测的图片的结果几乎相同，并且准确性很低（0.49）（附有图片）。 photo_of_the_results_from_sample_data

我尝试在adam，SGD，RMSprop之间更改优化器我试图将损失更改为稀疏类别我试图将指标更改为混淆矩阵（由于某种原因，它甚至无法识别）没有任何作用......

有人可以帮我修复我的代码吗？非常感谢您！

from keras.models import Sequential
from keras.layers import Dense, Dropout, Flatten, Conv2D, 
MaxPooling2D
from keras.optimizers import adam, SGD, RMSprop
from keras.preprocessing.image import ImageDataGenerator

# 1 - Load data
train_dataset = "\\Pictures_Dogs_DataSet\\Train1/"
validation_dataset="\\Pictures_Dogs_DataSet\\Validation1/"
test_dataset = "\\Pictures_Dogs_DataSet\\Test/"

# 2 - Create network layers
image_width = 200 
image_height = 200

model = Sequential()

# add convolutional layer (CNN)
# 32 X 32 pixel matrix
# 3 layars (RGB)
# 3 X 3 sliding matrix
# a pixel value for each colour is between 0-255
model.add(Conv2D(filters=32,kernel_size=(3,3),input_shape= 
(image_width,image_height,3),activation='relu'))
model.add(MaxPooling2D(pool_size=(2,2)))
# Add another block for more abstract features
model.add(Conv2D(32,(3,3), input_shape=(image_width, 
image_height,3),activation='relu'))
model.add(MaxPooling2D(pool_size=(2,2)))
# Add another block for more abstract features
model.add(Conv2D(64,(3,3), input_shape=(image_width, 
image_height,3),activation='relu'))
model.add(MaxPooling2D(pool_size=(2,2)))
# Add another block for more abstract features
model.add(Conv2D(128,(3,3), input_shape=(image_width, 
image_height,3),activation='relu'))
model.add(MaxPooling2D(pool_size=(2,2)))

# Let's DropOut some connections to avoid over-fitting.
model.add(Flatten())
model.add(Dense(128,activation='relu'))
model.add(Dropout(0.25))
model.add(Dense(32,activation='relu'))
model.add(Dropout(0.25))
model.add(Dense(1,activation='sigmoid'))


# Using the prediction model
sgd = SGD(lr=0.001)
model.compile(optimizer=sgd,loss='binary_crossentropy',
          metrics=['accuracy'])

# used to rescale the pixel values from [0, 255] to [0, 1] 
interval
datagen =ImageDataGenerator(rescale=1./255)

# automatically retrieve images and their classes for train 
and validation sets
train_generator = datagen.flow_from_directory(
        train_dataset,
        target_size=(image_width, image_height),
        batch_size=32,
        class_mode='binary')

validation_generator = datagen.flow_from_directory(
        validation_dataset,
        target_size=(image_width, image_height),
        batch_size=32,
        class_mode='binary')

#  3 - Train
model.fit_generator(
        train_generator,
        steps_per_epoch=10,
        epochs=1,  
        #just 1 epoch to practice and make sure it works#
        validation_data=validation_generator,
        validation_steps=5)


#evaluate the accuracy of the model
model.evaluate_generator(validation_generator,verbose=0)

#save modules
#model.save('model.h5')

# 5 - predict
test_generator =datagen.flow_from_directory(
        directory=test_dataset,
        target_size=(image_width,image_height),
        class_mode=None,
        batch_size=32)

print(model.predict_generator(test_generator))

基于CNN模型的图像预测总是返回相似（和居中）的结果

0 个答案: