如何在多个输入中使用 fit_generator

问题描述

可以有两个 fit_generator 吗?

Is it possible to have two fit_generator?

我正在创建一个具有两个输入的模型，模型配置如下所示.

I'm creating a model with two inputs,The model configuration is shown below.

标签 Y 对 X1 和 X2 数据使用相同的标签.

Label Y uses the same labeling for X1 and X2 data.

以下错误将继续发生.

检查模型输入时出错:您传递给模型的 Numpy 数组列表不是模型预期的大小.预期的查看 2 个数组，但得到了以下 1 个数组的列表:[数组([[[[0.75686276, 0.75686276, 0.75686276],[0.75686276, 0.75686276, 0.75686276],[0.75686276, 0.75686276, 0.75686276],...,[0.65882355, 0.65882355, 0.65882355...

我的代码如下:

def generator_two_img(X1, X2, Y,batch_size):
    generator = ImageDataGenerator(rotation_range=15,
                                   width_shift_range=0.2,
                                   height_shift_range=0.2,
                                   shear_range=0.2,
                                   zoom_range=0.2,
                                   horizontal_flip=True,
                                   fill_mode='nearest')

    genX1 = generator.flow(X1, Y, batch_size=batch_size)
    genX2 = generator.flow(X2, Y, batch_size=batch_size)

    while True:
        X1 = genX1.__next__()
        X2 = genX2.__next__()
        yield [X1, X2], Y
  """
      .................................
  """
hist = model.fit_generator(generator_two_img(x_train, x_train_landmark,
                y_train, batch_size),
                steps_per_epoch=len(x_train) // batch_size, epochs=nb_epoch,
                callbacks = callbacks,
                validation_data=(x_validation, y_validation),
                validation_steps=x_validation.shape[0] // batch_size,
                `enter code here`verbose=1)

推荐答案

试试这个生成器:

def generator_two_img(X1, X2, y, batch_size):
    genX1 = gen.flow(X1, y,  batch_size=batch_size, seed=1)
    genX2 = gen.flow(X2, y, batch_size=batch_size, seed=1)
    while True:
        X1i = genX1.next()
        X2i = genX2.next()
        yield [X1i[0], X2i[0]], X1i[1]

3 个输入的生成器:

def generator_three_img(X1, X2, X3, y, batch_size):
    genX1 = gen.flow(X1, y,  batch_size=batch_size, seed=1)
    genX2 = gen.flow(X2, y, batch_size=batch_size, seed=1)
    genX3 = gen.flow(X3, y, batch_size=batch_size, seed=1)
    while True:
        X1i = genX1.next()
        X2i = genX2.next()
        X3i = genX3.next()
        yield [X1i[0], X2i[0], X3i[0]], X1i[1]

这篇关于如何在多个输入中使用 fit_generator的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持！