train_generator使用keras flow_from_directory(下面的示例代码(从磁盘读取RGB图像数据批次。但是就我而言,我有两个图像目录,使我想阅读一对图像并沿深度轴堆叠以形成6通道图像(即2x r,2x g,2x g,2x b通道(转到fit_generator。
那么,我的问题是如何在使用keras flow_from_directory?
的同时,沿深度轴沿深度轴组合两个RGB图像来准备6通道输入数据。我正在遵循此处的CNN代码:
https://gist.github.com/fchollet/0830affa1f7f19f19fd47b06d4cf89ed444d
train_generator = train_datagen.flow_from_directory(
train_data_dir,
target_size=(img_width, img_height),
batch_size=batch_size,
class_mode='binary')
flow_from_directory
返回迭代器。使用MAP可以使两个迭代器的输出连接。(未测试代码(
train_generator1 = train_datagen.flow_from_directory(
train_data_dir1,
target_size=(img_width, img_height),
batch_size=batch_size,
class_mode='binary')
train_generator2 = train_datagen.flow_from_directory(
train_data_dir2,
target_size=(img_width, img_height),
batch_size=batch_size,
class_mode='binary')
map(lambda x1, y1, x2, y2: tf.concat([x1,x2], axis=-1), tf.concat([y1,y2], axis=-1), train_generator1, train_generator2)