如果不再次训练，就无法使用张量流保存 CNN 模型

Question

我正在尝试保存一个Sequential CNN模型。我发现我可以使用model.save()保存它，但在我尝试使用keras.models.load_model()将它加载回来后，它又开始训练自己了。

我如何保存我的模型，这样我就不需要再次训练了？

此外，在训练期间，我收到了几次以下警告，上面写着：

/15 [=>............................] - ETA: 39s - loss: 0.6936 - accuracy: 0.50782022-10-11 
17:31:06.794142: W tensorflow/tsl/framework/cpu_allocator_impl.cc:82] 
Allocation of 358875136 exceeds 10% of free system memory.

这可能是原因吗？

以下是该型号的代码：

import tensorflow as tf
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Conv2D, Flatten, Dropout, MaxPooling2D
from tensorflow.keras.preprocessing.image import ImageDataGenerator
import os
import numpy as np
import matplotlib.pyplot as plt
import pickle
PATH = 'cats_and_dogscats_and_dogs'
train_dir = os.path.join(PATH, 'train')
validation_dir = os.path.join(PATH, 'validation')
test_dir = os.path.join(PATH, 'test')

# Variables for pre-processing and training.
batch_size = 128
epochs =1
IMG_HEIGHT = 150
IMG_WIDTH = 150

train_image_generator = ImageDataGenerator(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(batch_size=batch_size,
directory=train_dir,
target_size=(IMG_HEIGHT, IMG_WIDTH),
class_mode='binary')
val_data_gen = validation_image_generator.flow_from_directory(validation_dir, target_size=(IMG_HEIGHT, IMG_WIDTH),  class_mode='categorical',batch_size=batch_size)
test_data_gen = test_image_generator.flow_from_directory(test_dir, target_size=(IMG_HEIGHT, IMG_WIDTH), classes ='.',class_mode='categorical',  batch_size=batch_size, shuffle = False)
#I,ve found that you can use classes = ".", to get test data labels (labels when there are no subdirectories ))
from tensorflow.python.framework.func_graph import flatten

model = tf.keras.Sequential()   
model.add(tf.keras.layers.Conv2D(32, (3,3) , input_shape = (150,150,3)))
model.add(tf.keras.layers.MaxPool2D(pool_size=(2, 2)))
model.add(tf.keras.layers.Conv2D(64, (3,3),activation = 'relu'))
model.add(tf.keras.layers.MaxPool2D(pool_size=(2, 2)))
model.add(tf.keras.layers.Conv2D(64, (3,3),activation = 'relu'))
model.add(tf.keras.layers.Flatten())
model.add(tf.keras.layers.Dense(64))
model.add(tf.keras.layers.Dense(1,activation = 'sigmoid'))
model.compile(optimizer='adam',
loss=tf.keras.losses.BinaryCrossentropy(),
metrics=['accuracy'])
model.fit(train_data_gen,
epochs=epochs, 
batch_size = batch_size,
validation_data=val_data_gen,
steps_per_epoch =2000//batch_size, 
validation_steps=800//batch_size)

model.save('CatDog.h5')

和另一个文件的代码，我试图上传模型到：

import tensorflow as tf
import pandas
import tkinter
import os
from CNNmodel import IMG_HEIGHT, IMG_WIDTH
from tensorflow.keras.preprocessing.image import ImageDataGenerator #type: ignore
from tensorflow import keras
model = keras.models.load_model('CatDog.h5')```

Answer 1

它再次开始训练，因为您可能再次调用model.fit(...)。

这足以加载回一个模型：

from tensorflow import keras
model = keras.models.load_model('path/to/location')

如果你想获得预测，那么你会有这样的东西，无需再次训练：

prediction = model(test_data, training=False)

Answer 2

EDIT：实际上，我已经修复了它，感谢您的帮助，在我的第二个文件中，我从模型创建文件中导入了一些变量

from CNNmodel import IMG_HEIGHT, IMG_WIDTH

我想模型.它不知怎么又被称为