neural-network - 带有 Conv2D 层的 VAE 返回“InvalidArgumentError：不兼容的形状”

Question

import random
import os
from tensorflow.keras.layers import Flatten,Dense, Conv2D, MaxPooling2D, Input, UpSampling2D, Reshape
from tensorflow.keras.callbacks import EarlyStopping
import tensorflow as tf

seed_n=777
random.seed(seed_n)
tf.random.set_seed(seed_n)

ds_train = tf.keras.preprocessing.image_dataset_from_directory("./images/train",
                                                               labels='inferred',
                                                               #label_mode='int',
                                                               class_names=None,
                                                               color_mode='rgb',
                                                               image_size=(224,224),
                                                               shuffle=True,
                                                               seed=seed_n,
                                                               validation_split=None,
                                                               subset=None,
                                                               interpolation='bilinear',
                                                               follow_links=False
                                                            )

ds_test = tf.keras.preprocessing.image_dataset_from_directory("./images/test",
                                                              labels='inferred',
                                                              #label_mode=None,
                                                              class_names=None,
                                                              color_mode='rgb',
                                                              image_size=(224,224),
                                                              shuffle=True,
                                                              seed=seed_n,
                                                              validation_split=None,
                                                              subset=None,
                                                              interpolation='bilinear',
                                                              follow_links=False
                                                            )

def autoencoder(z_dim):
    # inputs = Input(shape=[224,224,3])
    inputs = Input((224,224,3))
    x = inputs
    x = Conv2D(filters=8, kernel_size=(3, 3), strides=2, padding="same", activation="relu")(x)
    x = Conv2D(filters=8, kernel_size=(3, 3), strides=1, padding="same", activation="relu")(x)
    x = Conv2D(filters=8, kernel_size=(3, 3), strides=2, padding="same", activation="relu")(x)
    x = Conv2D(filters=8, kernel_size=(3, 3), strides=1, padding="same", activation="relu")(x)
    x = Flatten()(x)
    x = Dense(z_dim, activation="relu")(x)
    x = Dense(7*7*64, activation="relu")(x)
    x = Reshape((7, 7, 64))(x)
    x = Conv2D(filters=64, kernel_size=(3, 3), strides=1, padding="same", activation="relu")(x)
    x = UpSampling2D((2, 2))(x)
    x = Conv2D(filters=32, kernel_size=(3, 3), strides=1, padding="same", activation="relu")(x)
    x = UpSampling2D((16, 16))(x)
    out = Conv2D(filters=3, kernel_size=(3, 3), strides=1, padding="same", activation="sigmoid")(x) # Filter value should be the number of colour channels

    return Model(inputs=inputs, outputs=out, name="autoencoder")

z_dim = 1000
autoencoder = autoencoder(z_dim)

autoencoder.compile(loss="mse", optimizer=tf.keras.optimizers.RMSprop(learning_rate=3e-4))

autoencoder.fit(ds_train, validation_data=ds_test, epochs=1,callbacks = [EarlyStopping(monitor="val_loss", patience=2)])

网络架构如下：

_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
input_2 (InputLayer)         [(None, 224, 224, 3)]     0         
_________________________________________________________________
conv2d_7 (Conv2D)            (None, 112, 112, 8)       224       
_________________________________________________________________
conv2d_8 (Conv2D)            (None, 112, 112, 8)       584       
_________________________________________________________________
conv2d_9 (Conv2D)            (None, 56, 56, 8)         584       
_________________________________________________________________
conv2d_10 (Conv2D)           (None, 56, 56, 8)         584       
_________________________________________________________________
flatten_1 (Flatten)          (None, 25088)             0         
_________________________________________________________________
dense_2 (Dense)              (None, 1000)              25089000  
_________________________________________________________________
dense_3 (Dense)              (None, 3136)              3139136   
_________________________________________________________________
reshape_1 (Reshape)          (None, 7, 7, 64)          0         
_________________________________________________________________
conv2d_11 (Conv2D)           (None, 7, 7, 64)          36928     
_________________________________________________________________
up_sampling2d_2 (UpSampling2 (None, 14, 14, 64)        0         
_________________________________________________________________
conv2d_12 (Conv2D)           (None, 14, 14, 32)        18464     
_________________________________________________________________
up_sampling2d_3 (UpSampling2 (None, 224, 224, 32)      0         
_________________________________________________________________
conv2d_13 (Conv2D)           (None, 224, 224, 3)       867       
=================================================================
Total params: 28,286,371
Trainable params: 28,286,371
Non-trainable params: 0
_________________________________________________________________

我收到以下错误：

InvalidArgumentError:  Incompatible shapes: [3,224,224,3] vs. [3,1]
     [[node gradient_tape/mean_squared_error/BroadcastGradientArgs (defined at <ipython-input-14-a033834688e2>:1) ]] [Op:__inference_train_function_3138]

Function call stack:
train_function

我怀疑错误出在批量大小或损失函数中。但是除非我设置batch_size = 1，否则我无法使其工作。模型摘要中的输入/输出形状对我来说似乎很好，但我显然遗漏了一些东西......

编辑：batch_size=1 有时有效，有时无效...只是尝试了一个更大的数据集（7730 个图像用于训练，499 个图像用于验证）并得到了不同的错误：

ValueError: No gradients provided for any variable: ['conv2d_7/kernel:0', 'conv2d_7/bias:0', 'conv2d_8/kernel:0', 'conv2d_8/bias:0', 'conv2d_9/kernel:0', 'conv2d_9/bias:0', 'conv2d_10/kernel:0', 'conv2d_10/bias:0', 'dense_2/kernel:0', 'dense_2/bias:0', 'dense_3/kernel:0', 'dense_3/bias:0', 'conv2d_11/kernel:0', 'conv2d_11/bias:0', 'conv2d_12/kernel:0', 'conv2d_12/bias:0', 'conv2d_13/kernel:0', 'conv2d_13/bias:0'].

标签： neural-networkconv-neural-networkcomputer-vision