从本地文件夹导入图像，而不是使用 Keras/Tensorflow 数据集

Question

Hi Can someone please help me to change this code so that it wont get data from keras mnist. instead it will be getting data from local folder. where do i need to make changes in it. and where in this code we can use shuffle = true.

# Credits: https://github.com/keras-team/keras/blob/master/examples/mnist_cnn.py

from __future__ import print_function
import keras
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 = 12

# input image dimensions
img_rows, img_cols = 28, 28

# the data, split between train and test sets
(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')

# convert class vectors to binary class matrices
y_train = keras.utils.to_categorical(y_train, num_classes)
y_test = keras.utils.to_categorical(y_test, num_classes)

model1 = Sequential()
model1.add(Conv2D(32, kernel_size=(3, 3),
                 activation='relu',
                 input_shape=input_shape))
model1.add(Conv2D(32, (3, 3), activation='relu'))
model1.add(MaxPooling2D(pool_size=(2, 2)))
model1.add(Conv2D(64, (3, 3), activation='relu'))
model1.add(Dropout(0.25))
model1.add(Flatten())
model1.add(Dense(128, activation='relu'))
model1.add(Dropout(0.5))
model1.add(Dense(num_classes, activation='softmax'))

model1.compile(loss=keras.losses.categorical_crossentropy,
              optimizer=keras.optimizers.Adadelta(),
              metrics=['accuracy'])

Answer 1

x = # some array containing image data
y = # some array containing labels

(x_train, x_test), (y_train, y_test) = sklearn.model_selection.train_test_split(x, y)  # shuffle=True by default

Note the change in ordering.

Answer 2

For doing what you want, You have multiple options. I write an approach with using flow_from_directory . For using this appraoch your data in local folder should be like this:

---

Loading images from path:

import tensorflow as tf

datagen = tf.keras.preprocessing.image.ImageDataGenerator(rescale=1./255)
train_generator = datagen.flow_from_directory(
        'data/train',
        target_size=(100, 100),
        batch_size=32,
        shuffle=True,
        class_mode='categorical')
test_generator = datagen.flow_from_directory(
        'data/test',
        target_size=(100, 100),
        batch_size=32,
        class_mode='categorical')

Output:

Found 30 images belonging to 3 classes.
Found 30 images belonging to 3 classes.

---

Train model after loading data: (accuracy decrease because images are random numbers and they are not real images .)

from tensorflow.keras import Sequential
from tensorflow.keras.layers import Conv2D, Flatten, Dense, Dropout, MaxPooling2D


model1 = Sequential()
model1.add(Conv2D(32, kernel_size=(3, 3),
                 activation='relu',
                 input_shape=(100,100,3)))
model1.add(Conv2D(32, (3, 3), activation='relu'))
model1.add(MaxPooling2D(pool_size=(2, 2)))
model1.add(Conv2D(64, (3, 3), activation='relu'))
model1.add(Dropout(0.25))
model1.add(Flatten())
model1.add(Dense(128, activation='relu'))
model1.add(Dropout(0.5))
model1.add(Dense(4, activation='softmax'))
model1.compile(loss=tf.keras.losses.categorical_crossentropy,
              optimizer=tf.keras.optimizers.Adadelta(),
              metrics=['accuracy'])

model1.fit(train_generator, epochs=2)
model1.evaluate(test_generator)

Output:

Epoch 1/2
1/1 [==============================] - 3s 3s/step - loss: 1.3012 - accuracy: 0.3667
Epoch 2/2
1/1 [==============================] - 2s 2s/step - loss: 1.3266 - accuracy: 0.3000
1/1 [==============================] - 1s 754ms/step - loss: 1.3075 - accuracy: 0.3333
[1.3074820041656494, 0.3333333432674408]

---

Generating random images in path like the image:

import numpy as np
from PIL import Image

for idx, loc in enumerate(['data/train', 'data/test']*10):
    for category in ['apple', 'orange', 'banana']:
        imarray = np.random.rand(100,100,3) * 255
        im = Image.fromarray(imarray.astype('uint8')).convert('RGB')
        im.save(f'{loc}/{category}/img_{idx}.png')