【问题标题】:keras CNN can't recognize images with one channelkeras CNN 无法识别只有一个通道的图像
【发布时间】:2019-07-03 05:25:22
【问题描述】:

我正在尝试在我的数据上训练CNN model,这是一组gray-scaleOpenCV 从numpy 数组生成的图像,图像是75*70 像素。我收到以下错误:

    ValueError: Error when checking input: expected conv2d_25_input to have 
    shape (64, 64, 1) but got array with shape (64, 64, 3) 

这是我的代码:

# Importing the Keras libraries and packages
from keras.models import Sequential
from keras.layers import Convolution2D
from keras.layers import MaxPooling2D
from keras.layers import Flatten
from keras.layers import Dense

# Initialising the CNN
classifier = Sequential()

# Step 1 - Convolution
classifier.add(Convolution2D(32, 3, 3, input_shape = (64,64,1), activation = 'relu'))

# Step 2 - Pooling
classifier.add(MaxPooling2D(pool_size = (2, 2)))

# Adding a second convolutional layer
#classifier.add(Convolution2D(32, 3, 3, activation = 'relu'))
#classifier.add(MaxPooling2D(pool_size = (2, 2)))

# Step 3 - Flattening
classifier.add(Flatten())

# Step 4 - Full connection
classifier.add(Dense(output_dim = 128, activation = 'relu'))
classifier.add(Dense(output_dim = 750, activation = 'softmax'))

# Compiling the CNN
classifier.compile(optimizer = 'adam', loss = 'categorical_crossentropy', metrics = ['accuracy'])

# Part 2 - Fitting the CNN to the images

from keras.preprocessing.image import ImageDataGenerator


train_datagen = ImageDataGenerator(rescale = 1./255)


test_datagen = ImageDataGenerator(rescale = 1./255)

training_set = train_datagen.flow_from_directory('train',
                                                 target_size = (64, 64),
                                                 batch_size = 32,
                                                 class_mode = 'categorical')

test_set = test_datagen.flow_from_directory('test',
                                            target_size = (64, 64),
                                            batch_size = 32,
                                            class_mode = 'categorical')


classifier.fit_generator(training_set,
                         samples_per_epoch = 525,
                         nb_epoch = 25,
                         validation_data = test_set,
                         nb_val_samples = 225)

我的图片只有一个通道,但仍然出现此输入形状错误, 谁能帮帮我?

编辑: 我在keras documentation找到答案,ImageDataGenerator默认color_modergb,所以我改成grayscale,解决了input shape的问题 代码如下所示;

training_set = train_datagen.flow_from_directory('train',
                                                 target_size = (64,64),
                                                 color_mode = 'grayscale',
                                                 batch_size = 32,
                                                 class_mode = 'categorical')

test_set = test_datagen.flow_from_directory('test',
                                            target_size = (64, 64),
                                            color_mode = 'grayscale',
                                            batch_size = 32,
                                            class_mode = 'categorical')

但是,我遇到了另一个错误: ValueError: Error when checking target: expected dense_87 to have shape (750,) but got array with shape (0,) 我想不通……!!

【问题讨论】:

  • 从你的数据中取出一张照片并打印出来,我认为你错过了一些灰度照片。再次尝试 cv2.imread(img_path, 0) 所有这些。
  • 问题是我应该在转换中提供一个输入形状。层,不是吗?
  • 如果您想处理灰度图片,请使用 cv2.cvt 将所有数据转换为 1 通道的灰度。如果您想使用原始数据,请将 input_shape 更改为 (64,64,3)
  • 我想处理灰度图像,实际上我是从二维数组生成这些图像的,所以每个像素只有一个值,所以我认为它们应该只有一个通道。此外,我将输入形状更改为 (64,64,3),出现另一个错误:ValueError: Error when checks target: expected dense_47 to have shape (750,) but got array with shape (0,)
  • classifier.add(Convolution2D(32, 3, 3, input_shape = (64,64,3), activation = 'relu', dim_ordering='tf')) 试试这个。

标签: python keras conv-neural-network


【解决方案1】:

您为什么不尝试使用 .reshape(64,64,1) 来确保您的图像在一个通道中

【讨论】:

  • 谢谢@Harshit Gupta,我应该这样做:train_datagen.reshape(64,64,1) ??
猜你喜欢
  • 1970-01-01
  • 1970-01-01
  • 2020-05-31
  • 1970-01-01
  • 1970-01-01
  • 1970-01-01
  • 1970-01-01
  • 2019-10-22
  • 1970-01-01
相关资源
最近更新 更多