尝试拟合模型时出错 - Tensorflow CNN

Question

我正在尝试创建一个用于图像分类（猫和狗）的 CNN。在我使用 fit 之前，一切都运行良好。由于我是初学者，我担心我没有正确创建我的顺序模型。我也不确定我每个时期的步骤和验证步骤是否正确。
使用 history = model.fit 时出现以下错误：

2022-07-12 12:10:04.374122: I tensorflow/compiler/mlir/mlir_graph_optimization_pass.cc:185] None of the MLIR Optimization Passes are enabled (registered 2)
Epoch 1/15
2022-07-12 11:57:57.465865: I tensorflow/compiler/mlir/mlir_graph_optimization_pass.cc:185] None of the MLIR Optimization Passes are enabled (registered 2)>
2022-07-12 11:57:58.808025: I tensorflow/stream_executor/cuda/cuda_dnn.cc:369] Loaded cuDNN version 8201
2022-07-12 11:58:00.536214: W tensorflow/core/framework/op_kernel.cc:1680] Invalid argument: required broadcastable shapes
2022-07-12 11:58:00.536457: W tensorflow/core/framework/op_kernel.cc:1680] Invalid argument: required broadcastable shapes
2022-07-12 11:58:00.536536: W tensorflow/core/framework/op_kernel.cc:1680] Invalid argument: required broadcastable shapes
Traceback (most recent call last):
  File "C:\Users\myPC\PycharmProjects\pythonProject\catsdogs.py", line 75, in <module>
    history = model.fit(train_data_gen, epochs=epochs, validation_data=val_data_gen, steps_per_epoch=int(np.ceil(train_data_gen.n / float(batch_size))), validation_steps=int(np.ceil(val_data_gen.n / float(batch_size))))
  File "C:\Users\myPC\anaconda3\envs\tensorflow\lib\site-packages\keras\engine\training.py", line 1184, in fit
    tmp_logs = self.train_function(iterator)
  File "C:\Users\myPC\anaconda3\envs\tensorflow\lib\site-packages\tensorflow\python\eager\def_function.py", line 885, in __call__
    result = self._call(*args, **kwds)
  File "C:\Users\myPC\anaconda3\envs\tensorflow\lib\site-packages\tensorflow\python\eager\def_function.py", line 950, in _call
    return self._stateless_fn(*args, **kwds)
  File "C:\Users\myPC\anaconda3\envs\tensorflow\lib\site-packages\tensorflow\python\eager\function.py", line 3039, in __call__
    return graph_function._call_flat(
  File "C:\Users\myPC\anaconda3\envs\tensorflow\lib\site-packages\tensorflow\python\eager\function.py", line 1963, in _call_flat
    return self._build_call_outputs(self._inference_function.call(
  File "C:\Users\myPC\anaconda3\envs\tensorflow\lib\site-packages\tensorflow\python\eager\function.py", line 591, in call
    outputs = execute.execute(
  File "C:\Users\myPC\anaconda3\envs\tensorflow\lib\site-packages\tensorflow\python\eager\execute.py", line 59, in quick_execute
    tensors = pywrap_tfe.TFE_Py_Execute(ctx._handle, device_name, op_name,
tensorflow.python.framework.errors_impl.InvalidArgumentError:  required broadcastable shapes
     [[node Equal (defined at \PycharmProjects\pythonProject\catsdogs.py:75) ]] [Op:__inference_train_function_733]

Function call stack:
train_function

2022-07-12 11:58:00.649539: W tensorflow/core/kernels/data/generator_dataset_op.cc:107] Error occurred when finalizing GeneratorDataset iterator: Failed precondition: Python interpreter state is not initialized. The process may be terminated.
     [[{{node PyFunc}}]]

Process finished with exit code 1>

这是我使用的代码（Tensorflow 版本是 2.6）：

import tensorflow as tf
import keras

from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Conv2D, Flatten, Dropout, MaxPool2D
from tensorflow.keras.preprocessing.image import ImageDataGenerator

import os
import numpy as np
import matplotlib.pyplot as plt

# GET PROJECT FILES
PATH = 'cats_and_dogs'

train_dir = os.path.join(PATH, 'train')
validation_dir = os.path.join(PATH, 'validation')
test_dir = os.path.join(PATH, 'test')

# GET NUMBER OF FILES IN EACH DIRECTORY
total_train = sum([len(files) for r, d, files in os.walk(train_dir)])
total_val = sum([len(files) for r, d, files in os.walk(validation_dir)])
total_test = len(os.listdir(test_dir))

# VARIABLES FOR PRE-PROCESSING AND TRAINING.
batch_size = 128
epochs = 15
IMG_HEIGHT = 150
IMG_WIDTH = 150

# CREATE IMAGE DATA GENERATORS
train_image_generator = ImageDataGenerator(rotation_range=0.5, zoom_range=0.2, horizontal_flip=True, vertical_flip=True,
                                           rescale=1. / 255)
validation_image_generator = ImageDataGenerator(rescale=1. / 255)
test_image_generator = ImageDataGenerator(rescale=1. / 255)

train_data_gen = train_image_generator.flow_from_directory(directory=train_dir, target_size=(IMG_HEIGHT, IMG_WIDTH),
                                                           class_mode='binary', batch_size=batch_size)
val_data_gen = validation_image_generator.flow_from_directory(directory=validation_dir,
                                                              target_size=(IMG_HEIGHT, IMG_WIDTH),
                                                              class_mode='binary', batch_size=batch_size)
test_data_gen = test_image_generator.flow_from_directory(directory=test_dir, target_size=(IMG_HEIGHT, IMG_WIDTH),
                                                         class_mode='binary', batch_size=batch_size,
                                                         shuffle=False)


# CREATE MODEL
model = Sequential(
    [
        Conv2D(32, (3, 3), input_shape=(IMG_WIDTH, IMG_HEIGHT, 3)),
        MaxPool2D((2, 2)),
        Dense(1, activation='relu')
    ]
)

model.compile(optimizer=tf.keras.optimizers.Adam(learning_rate=1e-3),
              loss=tf.keras.losses.BinaryCrossentropy(),
              metrics=['accuracy'])

model.summary()
history = model.fit(train_data_gen, epochs=epochs, validation_data=val_data_gen, steps_per_epoch=int(np.ceil(train_data_gen.n / float(batch_size))), validation_steps=int(np.ceil(val_data_gen.n / float(batch_size))))

我觉得required broadcastable shapes 是罪魁祸首，但我可能错了。

Answer 1

经过一番研究，我终于让它在 pyCharm 中也能正常工作。问题是我没有使用 Flatten。这是我更改的代码。它现在工作正常。我仍然不确定为什么它不会在 Colab 中给我一个错误。

model = Sequential(
    [
        keras.Input(shape=(IMG_WIDTH, IMG_HEIGHT, 3)),
        Conv2D(32, 3, padding='same', activation='relu'),
        MaxPooling2D(),
        Conv2D(64, 3, activation='relu'),
        MaxPooling2D(),
        Conv2D(128, 3, activation='relu'),
        Flatten(),
        Dense(64, activation='relu'),
        Dropout(0.5),
        Dense(1, activation='sigmoid')
    ]
)

model.compile(
    optimizer=tf.keras.optimizers.Adam(learning_rate=3e-4),
    loss=keras.losses.BinaryCrossentropy(),
    metrics=["accuracy"]
)

model.summary()
history = model.fit(train_data_gen, epochs=epochs, validation_data=val_data_gen, batch_size=batch_size, verbose=2)