keras如何创建自己的generator

keras.utils.Sequence 是一个基类

class Sequence(object):
    """Base object for fitting to a sequence of data, such as a dataset.
    Every `Sequence` must implement the `__getitem__` and the `__len__` methods.
    If you want to modify your dataset between epochs you may implement
    `on_epoch_end`. The method `__getitem__` should return a complete batch.
    # Notes
    `Sequence` are a safer way to do multiprocessing. This structure guarantees
    that the network will only train once on each sample per epoch which is not
    the case with generators.
    # Examples
    ```python
        from skimage.io import imread
        from skimage.transform import resize
        import numpy as np
        # Here, `x_set` is list of path to the images
        # and `y_set` are the associated classes.
        class CIFAR10Sequence(Sequence):
            def __init__(self, x_set, y_set, batch_size):
                self.x, self.y = x_set, y_set
                self.batch_size = batch_size
            def __len__(self):
                return int(np.ceil(len(self.x) / float(self.batch_size)))
            def __getitem__(self, idx):
                batch_x = self.x[idx * self.batch_size:(idx + 1) * self.batch_size]
                batch_y = self.y[idx * self.batch_size:(idx + 1) * self.batch_size]
                return np.array([
                    resize(imread(file_name), (200, 200))
                       for file_name in batch_x]), np.array(batch_y)
    ```
    """

    use_sequence_api = True

    @abstractmethod
    def __getitem__(self, index):
        """Gets batch at position `index`.
        # Arguments
            index: position of the batch in the Sequence.
        # Returns
            A batch
        """
        raise NotImplementedError

    @abstractmethod
    def __len__(self):
        """Number of batch in the Sequence.
        # Returns
            The number of batches in the Sequence.
        """
        raise NotImplementedError

    def on_epoch_end(self):
        """Method called at the end of every epoch.
        """
        pass

    def __iter__(self):
        """Create a generator that iterate over the Sequence."""
        for item in (self[i] for i in range(len(self))):
            yield item

 所以我们需要去实现他的方法__len__,__getitem__

import keras
import numpy as np
import cv2


class DataGenerator(keras.utils.Sequence):
    def __init__(self,img_files=None, labels=None,batch_size=32,n_classes=11,shuffle=True,dim=(224,224,3)):
        self.dim = dim
        self.batch_size = batch_size
        self.img_files = img_files
        self.labels = labels
        self.n_classes = n_classes
        self.shuffle = shuffle
        self.cls2id={item:i for i,item  in enumerate(sorted(list(set(self.labels))))}
        self.on_epoch_end()
        

从keras.utils.Sequence类继承，__init__函数主要负责初始化img_files，labels等

其中self.on_epoch_end()主要用来初始化index,用于generator后续索引

    def on_epoch_end(self):
        self.indexes = np.arange(len(self.labels))
        if self.shuffle == True:
            np.random.shuffle(self.indexes)

__getitem__用来得到每一个batch的数据 

    def __getitem__(self,index):
        indexes = self.indexes[index*self.batch_size:(index+1)*self.batch_size]
        X, y =self.__data_generation(indexes)
        return X,y

__data_generation主要是迭代读入图片，对图片预处理

    def __data_generation(self,list_IDs_temp):
        X = np.empty((self.batch_size,*self.dim))
        y = np.empty((self.batch_size),dtype=int)
        for i, ID in enumerate(list_IDs_temp):
            img = cv2.imread(self.img_files[ID]).astype('float')
            label = self.labels[ID]
            img = cv2.resize(img,(224,224))
            img = img/255
            X[i,] = img
            y[i] = self.cls2id[label]
        return X,keras.utils.to_categorical(y,num_classes=self.n_classes)

完整训练代码如下

import keras
import numpy as np
import cv2


class DataGenerator(keras.utils.Sequence):
    def __init__(self,img_files=None, labels=None,batch_size=32,n_classes=11,shuffle=True,dim=(224,224,3)):
        self.dim = dim
        self.batch_size = batch_size
        self.img_files = img_files
        self.labels = labels
        self.n_classes = n_classes
        self.shuffle = shuffle
        self.cls2id={item:i for i,item  in enumerate(sorted(list(set(self.labels))))}
        self.on_epoch_end()
        
    def __len__(self):
        return int(len(self.img_files)/self.batch_size)
    
    def on_epoch_end(self):
        self.indexes = np.arange(len(self.labels))
        if self.shuffle == True:
            np.random.shuffle(self.indexes)
            
    def __getitem__(self,index):
        indexes = self.indexes[index*self.batch_size:(index+1)*self.batch_size]
        X, y =self.__data_generation(indexes)
        return X,y
        
    def __data_generation(self,list_IDs_temp):
        X = np.empty((self.batch_size,*self.dim))
        y = np.empty((self.batch_size),dtype=int)
        for i, ID in enumerate(list_IDs_temp):
            img = cv2.imread(self.img_files[ID]).astype('float')
            label = self.labels[ID]
            img = cv2.resize(img,(224,224))
            img = img/255
            X[i,] = img
            y[i] = self.cls2id[label]
        return X,keras.utils.to_categorical(y,num_classes=self.n_classes)


import osimport os
files = []
ids = []
for item in os.listdir('dog_classes/train/'):
    for dog in os.listdir('dog_classes/train/'+item):
        files.append('dog_classes/train/'+item+'/'+dog)
        ids.append(item)
ids = []
for item in os.listdir('dog_classes/train/'):
    for dog in os.listdir('dog_classes/train/'+item):
        files.append('dog_classes/train/'+item+'/'+dog)
        ids.append(item)

from keras.applications import ResNet50
from keras.models import Sequential
from keras.layers import Dense, Flatten, GlobalAveragePooling2D, Activation, Flatten, Dropout, BatchNormalization
from keras.callbacks import EarlyStopping, ModelCheckpoint, ReduceLROnPlateau
import h5py
##构建模型，用imagenet初始化
resnet_weights_path='resnet50_weights_tf_dim_ordering_tf_kernels_notop.h5'
my_new_model = Sequential()
my_new_model.add(ResNet50(include_top=False, pooling='avg', weights=resnet_weights_path))
my_new_model.add(Dense(512))
my_new_model.add(Activation('relu'))
my_new_model.add(Dropout(0.5))
my_new_model.add(Dense(3, activation='softmax'))


my_new_model.compile(optimizer='sgd', loss='categorical_crossentropy', metrics=['accuracy'])

from tensorflow.python.keras.applications.resnet50 import preprocess_input
from tensorflow.python.keras.preprocessing.image import ImageDataGenerator

image_size = 224
bs = 32
resnet_weights_path='resnet50_weights_tf_dim_ordering_tf_kernels_notop.h5'
my_new_model = Sequential()
my_new_model.add(ResNet50(include_top=False, pooling='avg', weights=resnet_weights_path))
my_new_model.add(Dense(512))
my_new_model.add(Activation('relu'))
my_new_model.add(Dropout(0.5))
my_new_model.add(Dense(11, activation='softmax'))


my_new_model.compile(optimizer='sgd', loss='categorical_crossentropy', metrics=['accuracy'])

my_new_model.fit_generator(generator=cls_gen,
                           steps_per_epoch=40,
                           epochs=50                      
                            )

my_new_model.fit_generator(generator=cls_gen,
                           steps_per_epoch=40,
                           epochs=50                      
                            )