Question

我正在尝试使用来自Keras的ImageDataGenerator和.flow（），以便为CNN分段任务实时增强图像数据。我对此类编程的基本理解提前道歉，因为我对它很陌生。我有一个功能如下，旨在增加图像数据和相应的数据掩码：

def image_augmentation(imgs, masks): 
    #  create two instances with the same arguments
    # create dictionary with the input augmentation values
    data_gen_args = dict(featurewise_center=False,
                         featurewise_std_normalization=False,
                         rotation_range=90.,
                         width_shift_range=0.1,
                         height_shift_range=0.1,
                         zoom_range=0.2, 
                         horizontal_flip=True,
                         vertical_flip = True)
    ## use this method with both images and masks
    image_datagen = ImageDataGenerator(**data_gen_args)
    mask_datagen = ImageDataGenerator(**data_gen_args)

    # Provide the same seed and keyword arguments to the fit and flow methods
    seed = 1
    ## fit the augmentation model to the images and masks with the same seed
    image_datagen.fit(imgs, augment=True, seed=seed)
    mask_datagen.fit(masks, augment=True, seed=seed)
    ## set the parameters for the data to come from (images)
    image_generator = image_datagen.flow(
        imgs,
        batch_size=32,
        shuffle=True,
        seed=seed)
    ## set the parameters for the data to come from (masks)
    mask_generator = mask_datagen.flow(
        masks,
        batch_size=32,
        shuffle=True,
        seed=seed)

    # combine generators into one which yields image and masks
    train_generator = zip(image_generator, mask_generator)
    ## return the train generator for input in the CNN 
    return train_generator

根据我的理解，火车发电机应该由32个增强图像和相应的掩模组成。但是，当我跑：

train_generator = image_augmentation(imgs_train, imgs_spine_mask_train)
        ## run the fit generator CNN
model.fit_generator(train_generator,steps_per_epoch=100,epochs=5000, callbacks=[model_checkpoint])

image_augmentation只会永远运行..没有错误

imgs_train，imgs_spine_mask_train都是4d数组

有谁能让我知道我在这里做错了什么？值得注意的是，我完全是stackoverflow的新手，所以如果我可以更改格式等，请告诉我。

干杯，

CNN的Keras图像和掩模增强

0 个答案: