当我复制您的模型时，在 resnet（即 Base_net）和密集层之间添加了 flatten 层，如下所示
Base_net = ResNet50(include_top = False, weights = 'imagenet', input_shape = input_shape)
model = Sequential()
model.add(Base_net)
model.add(Flatten())
model.add(Dense(units = num_classes, activation = 'softmax'))
model.summary()
model.compile(loss = 'categorical_crossentropy',
optimizer = 'Adam',
metrics = ['accuracy'])
输出：
Layer (type) Output Shape Param #
=================================================================
resnet50 (Model) (None, 7, 7, 2048) 23587712
_________________________________________________________________
flatten_5 (Flatten) (None, 100352) 0
_________________________________________________________________
dense_9 (Dense) (None, 1) 100353
=================================================================
Total params: 23,688,065
Trainable params: 23,634,945
Non-trainable params: 53,120
_________________________________________________________________
Epoch 1/3
100/100 [==============================] - 84s 837ms/step - loss: 0.0000e+00 - accuracy: 0.5085 - val_loss: 0.0000e+00 - val_accuracy: 0.5000
Epoch 2/3
100/100 [==============================] - 81s 806ms/step - loss: 0.0000e+00 - accuracy: 0.4953 - val_loss: 0.0000e+00 - val_accuracy: 0.5050
Epoch 3/3
100/100 [==============================] - 80s 801ms/step - loss: 0.0000e+00 - accuracy: 0.4965 - val_loss: 0.0000e+00 - val_accuracy: 0.5080
要解决您的问题，您必须将损失函数更改为并用最后一个密集层binary_crossentropy替换激活函数，从而解决您的问题。sigmoid为了获得更好的准确度值，请实施超参数调整。
注意：我们通常观察到批量大小是 2 的幂，这是因为优化的矩阵运算库的有效工作
请参考下面的工作代码，其中准确性
%tensorflow_version 2.x
import tensorflow as tf
from tensorflow.keras.applications import ResNet50
import os
import numpy as np
from keras.preprocessing.image import ImageDataGenerator
from tensorflow.keras.layers import Dense, Dropout, Flatten
from tensorflow.keras.optimizers import Adam
from tensorflow.keras.models import Sequential
from google.colab import drive
drive.mount('/content/drive')
train_dir = '/content/drive/My Drive/Dogs_Vs_Cats/train'
test_dir = '/content/drive/My Drive/Dogs_Vs_Cats/test'
img_width, img_height = 224, 224
input_shape = (img_width, img_height, 3)
#train_samples = 2000
#test_samples = 1000
epochs = 8
batch_size = 32
train_datagen = ImageDataGenerator(
rescale = 1. /255,
rotation_range=40,
width_shift_range=0.2,
height_shift_range=0.2,
shear_range = 0.2,
zoom_range = 0.2,
horizontal_flip = True)
test_datagen = ImageDataGenerator(
rescale = 1. /255)
train_data = train_datagen.flow_from_directory(
train_dir,
target_size = (img_width, img_height),
batch_size = batch_size,
class_mode = 'binary')
test_data = test_datagen.flow_from_directory(
test_dir,
target_size = (img_width, img_height),
#batch_size = batch_size,
class_mode = 'binary')
# The Convolutional Base of the Pre-Trained Model will be added as a Layer in this Model
Conv_Base = ResNet50(include_top = False, weights = 'imagenet', input_shape = input_shape)
model = Sequential()
model.add(Conv_Base)
model.add(Flatten())
model.add(Dense(units = 256, activation = 'relu'))
model.add(Dropout(0.5))
model.add(Dense(units = 1, activation = 'sigmoid'))
model.summary()
model.compile(loss = 'binary_crossentropy',
optimizer = 'Adam',
metrics = ['accuracy'])
model.fit_generator(
train_data,
steps_per_epoch = 100,
epochs = 8,
validation_data = test_data,
verbose = 1,
validation_steps = 32)
输出：
Found 2000 images belonging to 2 classes.
Found 1000 images belonging to 2 classes.
Model: "sequential_4"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
resnet50 (Model) (None, 7, 7, 2048) 23587712
_________________________________________________________________
flatten_4 (Flatten) (None, 100352) 0
_________________________________________________________________
dense_8 (Dense) (None, 256) 25690368
_________________________________________________________________
dropout_4 (Dropout) (None, 256) 0
_________________________________________________________________
dense_9 (Dense) (None, 1) 257
=================================================================
Total params: 49,278,337
Trainable params: 49,225,217
Non-trainable params: 53,120
_________________________________________________________________
Epoch 1/8
100/100 [==============================] - 1828s 18s/step - loss: 1.9865 - accuracy: 0.6401 - val_loss: 0.7398 - val_accuracy: 0.5000
Epoch 2/8
100/100 [==============================] - 65s 646ms/step - loss: 0.6642 - accuracy: 0.6951 - val_loss: 0.7094 - val_accuracy: 0.5000
Epoch 3/8
100/100 [==============================] - 65s 652ms/step - loss: 0.5662 - accuracy: 0.7673 - val_loss: 0.7088 - val_accuracy: 0.4950
Epoch 4/8
100/100 [==============================] - 65s 647ms/step - loss: 0.4362 - accuracy: 0.8330 - val_loss: 0.7026 - val_accuracy: 0.4950
Epoch 5/8
100/100 [==============================] - 65s 649ms/step - loss: 0.3998 - accuracy: 0.8405 - val_loss: 0.8281 - val_accuracy: 0.5070
Epoch 6/8
100/100 [==============================] - 65s 647ms/step - loss: 0.5733 - accuracy: 0.7986 - val_loss: 0.8699 - val_accuracy: 0.4970
Epoch 7/8
100/100 [==============================] - 65s 647ms/step - loss: 0.6359 - accuracy: 0.7860 - val_loss: 0.7071 - val_accuracy: 0.5020
Epoch 8/8
100/100 [==============================] - 65s 653ms/step - loss: 0.5514 - accuracy: 0.7673 - val_loss: 0.6590 - val_accuracy: 0.5910