python - 检查输入时出错：预期 conv2d_3_input 的形状为 (64, 64, 3) 但得到的数组的形状为 (64, 64, 1)

Question

基本上，我在学习 udemy 的机器学习 AZ 教程，在那里我学会了训练我自己的卷积神经网络模型，但是没有向我展示如何使用该训练过的模型——如何提供输入并获得输出。我在 youtube 上查看了一些教程，以通过 Keras 中的 cnn 传递单个图像，并阅读了本教程。当我运行代码时出现错误

ValueError：检查输入时出错：预期 conv2d_3_input 的形状为 (64, 64, 3) 但得到的数组的形状为 (64, 64, 1)

现在我完全不知道如何处理这个问题。如果有人能告诉我在我训练了我的网络之后该怎么做以及如何解决整个问题，我将非常感激我尝试改变 img_array = cv2.imread(filepath, cv2.IMREAD_GRAYSCALE) 阅读颜色，但我认为这与此无关

我用来训练卷积网络的代码：

import keras
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

初始化 cnn classifier = Sequential()

step1-应用卷积

classifier.add(Convolution2D(32,3,3, border_mode = 'same', input_shape=(64,64,3), activation = 'relu'))

应用最大池化

classifier.add(MaxPooling2D(pool_size=(2,2)))

添加另一个卷积层以获得更好的准确性

classifier.add(Convolution2D(32,3,3, border_mode = 'same', activation = 'relu'))
classifier.add(MaxPooling2D(pool_size=(2,2)))

应用展平

classifier.add(Flatten())

第 4 步 - 完全连接

classifier.add(Dense(output_dim = 128, activation = 'relu'))
#input
classifier.add(Dense(output_dim = 1, activation = 'sigmoid'))
#output

全部编译

classifier.compile(optimizer = 'adam', loss = 'binary_crossentropy', metrics = ['accuracy'])
#fitting the cnn to the images
from keras.preprocessing.image import ImageDataGenerator
#pixels take values between 0 & 255
train_datagen = ImageDataGenerator(rescale=1./255,                        
    shear_range=0.2,
    zoom_range=0.2,
    horizontal_flip=True)
test_datagen = ImageDataGenerator(rescale=1./255)

training_set = train_datagen.flow_from_directory('dataset/training_set', target_size=(64, 64), batch_size=32, class_mode='binary')       
test_set = test_datagen.flow_from_directory('dataset/test_set',
                                         target_size=(64, 64),
                                         batch_size=32,
                                         class_mode='binary')

classifier.fit_generator(training_set,
                     samples_per_epoch = 8000,
                     nb_epoch = 25,
                     verbose = 1,
                     validation_data = test_set,
                     nb_val_samples = 2000)

classifier.save('my_model.h5')

---

我在新文件中使用的代码来试用我训练有素的网络

import cv2
import tensorflow as tf
import keras
from keras.models import Sequential

Categories = ["Dogs","cats"]

def prepare(filepath):
    img_size = 64
    img_array = cv2.imread(filepath, cv2.IMREAD_GRAYSCALE)
    new_array = cv2.resize(img_array, (img_size,img_size))
    return new_array.reshape(-1,img_size,img_size, 1)
    classifier = keras.models.load_model('my_model.h5')

prediction = classifier.predict([prepare('/Users/m.zain/Documents/machine learning-2019 dps/Machine Learning A-Z Template Folder/Part 8 - Deep Learning/Section 40 - Convolutional Neural Networks (CNN)/dog.jpg')])
print(prediction)

prediction = classifier.predict([prepare('/Users/m.zain/Documents/machine learning-2019 dps/Machine Learning A-Z Template Folder/Part 8 - Deep Learning/Section 40 - Convolutional Neural Networks (CNN)/dog.jpg')])
 print(Categories[int(prediction[0][0])])

prediction = classifier.predict([prepare('/Users/m.zain/Documents/machine learning-2019 dps/Machine Learning A-Z Template Folder/Part 8 - Deep Learning/Section 40 - Convolutional Neural Networks (CNN)/cat.jpg')])
 print(Categories[int(prediction[0][0])])

---

当我执行代码时，我希望得到“狗”作为输出：

prediction = classifier.predict([prepare('/Users/m.zain/Documents/machine learning-2019 dps/Machine Learning A-Z Template Folder/Part 8 - Deep Learning/Section 40 - Convolutional Neural Networks (CNN)/dog.jpg')])
print(Categories[int(prediction[0][0])])

当我运行此命令时，将 cat 作为输出：

prediction = classifier.predict([prepare('/Users/m.zain/Documents/machine learning-2019 dps/Machine Learning A-Z Template Folder/Part 8 - Deep Learning/Section 40 - Convolutional Neural Networks (CNN)/cat.jpg')])
print(Categories[int(prediction[0][0])])

标签： pythonpython-3.xtensorflowkerasneural-network