检查输入时出错：预期 conv2d_1_input 的形状为 (28, 28, 1) 但得到的数组的形状为 (3, 224, 224)

Question

我该如何解决？ 我使用的代码显示在下面

这是用于将图像转换为矢量

import cv2
import numpy as np

file = cv2.imread('17316.png')
file = cv2.resize(file, (224, 224))
file = cv2.cvtColor(file, cv2.COLOR_BGR2RGB)
file = np.array(file).reshape((1, 3, 224, 224))
print(file.shape[0])

这是我应用的卷积神经网络的一部分，导致该错误我该怎么办，我该如何解决请建议我更改代码，以便我可以对我的数据集进行正确的预测？

model = Sequential()
model.add(Conv2D(32, kernel_size=(3, 3),
                 activation='relu',
                 input_shape=input_shape))
model.add(Conv2D(64, (3, 3), activation='relu'))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Dropout(0.25))
model.add(Flatten())
model.add(Dense(128, activation='relu'))
model.add(Dropout(0.5))
model.add(Dense(num_classes, activation='softmax'))

Answer 1

从错误消息中可以明显看出，模型需要 (28,28,1) 的图像形状。因此，在将图像输入模型之前尝试调整图像大小。

file = cv2.imread('17316.png')
file = cv2.resize(file, (28, 28))
file = cv2.cvtColor(file, cv2.COLOR_BGR2GRAY)
file = file.reshape((-1, 28, 28,1))

这将解决问题。

Answer 2

能否发布您的完整代码？ input_shape 的值是多少？我认为您应该将其设置为 (3, 224, 224)。显然，您的 data_format 是 channels_first，根据Keras conv2d documentation 默认是 channels_last。所以，我建议你使用你的第一个卷积层

model.add(Conv2D(32, kernel_size = (3, 3),
                 activation = 'relu',
                 input_shape = (3, 224, 224), 
                 data_format = "channels_first")

更新：根据您的代码，以下应该可以工作，但可能不会产生您想要的结果。您正在训练 mnist 数据集，该数据集需要 28x28x1 格式的图像，因此您必须调整大小，如 Mitiku 的答案所示。我希望这会有所帮助。

import keras
import cv2 
import numpy as np 
from keras.datasets import mnist
from keras.models import Sequential
from keras.layers import Dense, Dropout, Flatten 
from keras.layers import Conv2D, MaxPooling2D 
from keras import backend as K 

batch_size = 128 
num_classes = 10 
epochs = 1 
img_rows, img_cols = 28, 28
(x_train, y_train), (x_test, y_test) = mnist.load_data() 

if K.image_data_format() == 'channels_first': 
    x_train = x_train.reshape(x_train.shape[0], 1, img_rows, img_cols) 
    x_test = x_test.reshape(x_test.shape[0], 1, img_rows, img_cols) 
    input_shape = (1, img_rows, img_cols) 
else: 
    x_train = x_train.reshape(x_train.shape[0], img_rows, img_cols, 1) 
    x_test = x_test.reshape(x_test.shape[0], img_rows, img_cols, 1) 
    input_shape = (img_rows, img_cols, 1) 
    x_train = x_train.astype('float32')

x_test = x_test.astype('float32') 
x_train /= 255 
x_test /= 255 
print('x_train shape:', x_train.shape) 
print(x_train.shape[0], 'train samples') 
print(x_test.shape[0], 'test samples') 
y_train = keras.utils.to_categorical(y_train, num_classes) 
y_test = keras.utils.to_categorical(y_test, num_classes)

model = Sequential() 
model.add(Conv2D(32, kernel_size=(3, 3), activation='relu',\
                 input_shape = input_shape))
model.add(Conv2D(64, (3, 3), activation='relu')) 
model.add(MaxPooling2D(pool_size=(2, 2))) 
model.add(Dropout(0.25)) 
model.add(Flatten()) 
model.add(Dense(128, activation='relu')) 
model.add(Dropout(0.5)) 
model.add(Dense(num_classes, activation='softmax')) 
model.compile(loss=keras.losses.categorical_crossentropy, \
              optimizer=keras.optimizers.Adadelta(), \
              metrics=['accuracy'])

model.fit(x_train, y_train, batch_size=batch_size, epochs=epochs, \
          verbose=1, validation_data=(x_test, y_test)) 
score = model.evaluate(x_test, y_test, verbose=0) 

print('Test loss:', score[0]) 
print('Test accuracy:', score[1]) 

file = cv2.imread('17316.png') 
file = cv2.resize(file, (28, 28))
file = cv2.cvtColor(file, cv2.COLOR_BGR2GRAY)
file = file.reshape((28, 28,1))

model.predict(np.expand_dims(file, axis = 0))

更新 2： mnist 数据集有 10 个类。你有一个二元分类问题。您的输出将您的图像分类为第 8 类，对应于数字 7，因为 mnist 数据集类是从 0 到 9 的数字。我们必须知道这些类是如何编码的 - 这是特定于问题的。在这种情况下，要返回您可以执行的数字：

prediction = model.predict(np.expand_dims(file, axis = 0))
prediction = np.squeeze(prediction)
index = np.where(prediction == 1)[0]
number = (index - 1).item()
print("predicted number for my image: ", number)

最后一行返回预测包含 1 的位置的索引，由于索引从 1 开始，您可以从索引中减去 1 以获得与您的图像对应的数字。