我正在处理的任务是多类分割(每个图像上有0-3个类(。我有一个可工作的U-Net模型，可以在小数据集上进行很好的训练，然后我增强了数据集，现在我有了近15k 512x512灰度图像。我自然遇到了没有足够硬件资源(RAM、GPU(的问题，所以我决定改用谷歌colab并使用ImageDataGenerator。到目前为止，我遇到过无法解决的问题。
InvalidArgumentError：Conv2DSlowBackpropInput：out_backprop的大小与computed:actual=16，computed=32 spatial_dim:2不匹配输入：64滤波器：2输出：16步幅：2扩张：1[[节点模型/conv2d_transpose_1/conv2d_transfuse(定义于/usr/local/lib/python3.7/dist packages/ceras/backend.py:5360(]][操作：__推理_训练_功能_3151]
对我来说，唯一的解释是我没有很好地使用生成器。我将数据结构化为：
path_to_dataset
│
└───images_dir
│ │
│ └─── images_subdir
│ │ img1.png
│ │ img2.png
│ │ ...
│
└───masks_dir
│ │
│ └─── masks_subdir
│ │ img1.png
│ │ img2.png
│ │ ...
子目录只是为了使ImageDataGenerator工作。
data_gen_args = dict(rescale=1./255,)
image_datagen = ImageDataGenerator(**data_gen_args)
mask_datagen = ImageDataGenerator(**data_gen_args)
# image_datagen.fit(images)
# mask_datagen.fit(masks)
# Provide the same seed and keyword arguments to the fit and flow methods
seed = 1
image_generator = image_datagen.flow_from_directory(
'/content/drive/MyDrive/DP/preprocess_images/images/final_ds/orig_folder/',
batch_size=16,
class_mode=None,
# color_mode='grayscale',
seed=seed)
mask_generator = mask_datagen.flow_from_directory(
'/content/drive/MyDrive/DP/preprocess_images/images/final_ds/seg_greyscale_folder/',
batch_size=16,
class_mode=None,
# color_mode='grayscale',
seed=seed)
# combine generators into one which yields image and masks
train_generator = zip(image_generator, mask_generator)
callbacks = [
ModelCheckpoint('unet_512.h5', verbose=1, save_best_only=True),
EarlyStopping(patience=5, monitor='val_loss'),
TensorBoard(log_dir='logs_unet512')
]
history = model.fit(train_generator,
verbose=1,
epochs=50,
callbacks=callbacks,
# class_weight=class_weights,
shuffle=False)
到目前为止，我还没有处理为验证数据创建数据生成器的问题，因为我甚至无法使这部分工作。
对于好奇的人来说，这是模型。
# IMG_HEIGHT=512, IMG_WIDTH=512, IMG_CHANNELS=1
inputs = Input((IMG_HEIGHT, IMG_WIDTH, IMG_CHANNELS))
s = inputs
# Contraction path
c1 = Conv2D(16, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(s)
c1 = Dropout(0.1)(c1)
c1 = Conv2D(16, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(c1)
p1 = MaxPooling2D((2, 2))(c1)
c2 = Conv2D(32, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(p1)
c2 = Dropout(0.1)(c2)
c2 = Conv2D(32, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(c2)
p2 = MaxPooling2D((2, 2))(c2)
c3 = Conv2D(64, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(p2)
c3 = Dropout(0.2)(c3)
c3 = Conv2D(64, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(c3)
p3 = MaxPooling2D((2, 2))(c3)
c4 = Conv2D(128, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(p3)
c4 = Dropout(0.2)(c4)
c4 = Conv2D(128, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(c4)
p4 = MaxPooling2D(pool_size=(2, 2))(c4)
c5 = Conv2D(256, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(p4)
c5 = Dropout(0.3)(c5)
c5 = Conv2D(256, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(c5)
# Expansive path
u6 = Conv2DTranspose(128, (2, 2), strides=(2, 2), padding='same')(c5)
u6 = concatenate([u6, c4])
c6 = Conv2D(128, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(u6)
c6 = Dropout(0.2)(c6)
c6 = Conv2D(128, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(c6)
u7 = Conv2DTranspose(64, (2, 2), strides=(2, 2), padding='same')(c6)
u7 = concatenate([u7, c3])
c7 = Conv2D(64, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(u7)
c7 = Dropout(0.2)(c7)
c7 = Conv2D(64, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(c7)
u8 = Conv2DTranspose(32, (2, 2), strides=(2, 2), padding='same')(c7)
u8 = concatenate([u8, c2])
c8 = Conv2D(32, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(u8)
c8 = Dropout(0.1)(c8)
c8 = Conv2D(32, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(c8)
u9 = Conv2DTranspose(16, (2, 2), strides=(2, 2), padding='same')(c8)
u9 = concatenate([u9, c1], axis=3)
c9 = Conv2D(16, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(u9)
c9 = Dropout(0.1)(c9)
c9 = Conv2D(16, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(c9)
# n_classes=4
outputs = Conv2D(n_classes, (1, 1), activation='softmax')(c9)
model = Model(inputs=[inputs], outputs=[outputs])
编辑：还计划增加过滤器的数量，到目前为止，我正在运行以前在我的个人笔记本电脑上运行的模型
我没有找到一种方法让它与在实现中构建的keras一起工作，但是自定义生成器可以做到这一点。看起来大多数任务都处理得很好，但希望有一天会添加多类语义分割