我一直在尝试训练一个keras模型，但是在第一个纪元开始时它一直陷于困境。最糟糕的是它不会引发任何错误。我正在使用GTX 1050TI进行训练
下面是我的代码示例：
import tensorflow as tf
import os
from tensorflow import keras
from keras_preprocessing.image import ImageDataGenerator
from keras_applications.xception import Xception
import matplotlib.pyplot as plt
train_dir='C:\\Users\\AYERHAN MSUGHTER\\Documents\\chest_xray\\train'
single='C:\\Users\\AYERHAN MSUGHTER\\Documents\\chest_xray\\train\\person_1_bacteria_1'
validation_dir='C:\\Users\\AYERHAN MSUGHTER\\Documents\\chest_xray\\validation'
test_dir='C:\\Users\\AYERHAN MSUGHTER\\Documents\\chest_xray\\test'
shape_size=71
training_batch_size=10
validation_batch_size=10
target_size=([71,71])
#this is the image data generator that will generate additonal data for us
train_datagen=ImageDataGenerator(rescale=1/255,rotation_range=5,width_shift_range=0.1,
height_shift_range=0.05,shear_range=0.1,
zoom_range=0.15,horizontal_flip=True,
vertical_flip= False,
fill_mode='reflect')
validation_test_datagen= ImageDataGenerator(rescale=1/255)
train_generator=train_datagen.flow_from_directory(directory=train_dir,classes=('NORMAL','PNEUMONIA'), batch_size=training_batch_size,target_size=target_size,class_mode='categorical',shuffle=True,seed=5566)
validation_generator=validation_test_datagen.flow_from_directory(directory=validation_dir,
classes=('NORMAL','PNEUMONIA'),
target_size=target_size,
batch_size=training_batch_size,
class_mode='categorical',
shuffle=True,
seed=5566)
test_generator=validation_test_datagen.flow_from_directory(directory=train_dir,
classes=['NORMAL','PNEUMONIA'],
target_size=(shape_size,shape_size),
batch_size=1,
class_mode='categorical',
shuffle=True,
seed=5566)
conv_base=tf.keras.applications.Xception(weights='imagenet',include_top=False,input_shape=(shape_size,shape_size,3))
for layer in conv_base.layers[:-4]:
layer.trainable = False
for layer in conv_base.layers:
print(layer,layer.trainable)
import os
layer_input=tf.keras.layers.InputLayer(input_shape=[shape_size,shape_size,3],dtype=tf.float32,name='input_layer')
layer_global_average_pooling_2d=tf.keras.layers.AveragePooling2D()
layer_dense1=tf.keras.layers.Dense(units=1024,activation='relu', name='fc1')
layer_dropout1=tf.keras.layers.Dropout(rate=0.3,name='dropout1')
layer_dense2=tf.keras.layers.Dense(units=512,activation='relu', name='fc2')
layer_dropout2=tf.keras.layers.Dropout(rate=0.3,name='dropout2')
layer_dense3=tf.keras.layers.Dense(units=2,activation='relu', name='fc3')
#this would be our model
model= tf.keras.Sequential([
layer_input,
conv_base,
layer_global_average_pooling_2d,
layer_dense1,
layer_dropout1,
layer_dense2,
layer_dropout2,
layer_dense3
])
training_step_size= (len(list(train_dir )) / training_batch_size )
validation_step_size= (len(list(validation_dir )) / training_batch_size )
print(validation_step_size)
weight_adjustment=len(list(os.path.join(train_dir, '//NORMAL//' ))) / len(list(os.path.join(train_dir, '/PNEUMONIA/', )))
model.compile(
loss = "binary_crossentropy" ,
optimizer = tf.keras.optimizers.RMSprop(lr=1e-5) ,
metrics = ["accuracy"]
)
history=model.fit_generator(
train_generator ,
steps_per_epoch = training_step_size ,
# class_weight = [1,weight_adjustment],
epochs = 30 ,
validation_data = validation_generator ,
validation_steps = 2,
verbose=1
)
这是输出
"C:\Users\AYERHAN MSUGHTER\Anaconda3\python.exe" "C:/Users/AYERHAN MSUGHTER/PycharmProjects/test/test.py"
Found 0 images belonging to 2 classes.
Found 0 images belonging to 2 classes.
Found 0 images belonging to 2 classes.
WARNING:tensorflow:From C:\Users\AYERHAN MSUGHTER\Anaconda3\lib\site-packages\tensorflow\python\ops\init_ops.py:1251: calling VarianceScaling.__init__ (from tensorflow.python.ops.init_ops) with dtype is deprecated and will be removed in a future version.
Instructions for updating:
Call initializer instance with the dtype argument instead of passing it to the constructor
2019-11-29 22:16:25.367761: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX AVX2
2019-11-29 22:16:25.370526: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library nvcuda.dll
2019-11-29 22:16:26.168159: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1640] Found device 0 with properties:
name: GeForce GTX 1050 Ti major: 6 minor: 1 memoryClockRate(GHz): 1.62
pciBusID: 0000:01:00.0
2019-11-29 22:16:26.168581: I tensorflow/stream_executor/platform/default/dlopen_checker_stub.cc:25] GPU libraries are statically linked, skip dlopen check.
2019-11-29 22:16:26.169445: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1763] Adding visible gpu devices: 0
2019-11-29 22:16:26.657806: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1181] Device interconnect StreamExecutor with strength 1 edge matrix:
2019-11-29 22:16:26.658125: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1187] 0
2019-11-29 22:16:26.658300: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1200] 0: N
2019-11-29 22:16:26.659042: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1326] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 3000 MB memory) -> physical GPU (device: 0, name: GeForce GTX 1050 Ti, pci bus id: 0000:01:00.0, compute capability: 6.1)
<tensorflow.python.keras.engine.input_layer.InputLayer object at 0x000001BCDBD9E748> False
<tensorflow.python.keras.layers.convolutional.Conv2D object at 0x000001BCDD8F5198> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDD8F5F60> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCD6911C50> False
<tensorflow.python.keras.layers.convolutional.Conv2D object at 0x000001BCD6911710> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCD69D96D8> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCD69D9DA0> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCD6A663C8> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCD6ABE208> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCD6AE6160> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCD6AF82B0> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCD6B719B0> False
<tensorflow.python.keras.layers.convolutional.Conv2D object at 0x000001BCD69E5240> False
<tensorflow.python.keras.layers.pooling.MaxPooling2D object at 0x000001BCDDE11198> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCD6A5B780> False
<tensorflow.python.keras.layers.merge.Add object at 0x000001BCDDE2FA58> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCDDEA5828> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCDDEA5860> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDDF28550> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCDDF32DD8> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCDDF49F60> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDDFBC978> False
<tensorflow.python.keras.layers.convolutional.Conv2D object at 0x000001BCDDE4E438> False
<tensorflow.python.keras.layers.pooling.MaxPooling2D object at 0x000001BCDDFCB160> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDDEA5048> False
<tensorflow.python.keras.layers.merge.Add object at 0x000001BCDDFEA0B8> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCDE05DDD8> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCDE05D978> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDE0E0470> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCDE0E0BA8> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCDE0FCDA0> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDE172860> False
<tensorflow.python.keras.layers.convolutional.Conv2D object at 0x000001BCDE009B38> False
<tensorflow.python.keras.layers.pooling.MaxPooling2D object at 0x000001BCDE17F208> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDE05D128> False
<tensorflow.python.keras.layers.merge.Add object at 0x000001BCDE1A46A0> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCDE1C2C88> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCDE1EE7B8> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDE218550> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCDE21FE10> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCDE230FD0> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDE2A8A90> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCDE2B5278> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCDE2C2908> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDE33DEB8> False
<tensorflow.python.keras.layers.merge.Add object at 0x000001BCDE34F550> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCDE1C2588> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCDE367F28> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDE3E6748> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCDE3EF0F0> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCDE3FFC88> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDE474BA8> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCDE485390> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCDE48EA20> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDE520048> False
<tensorflow.python.keras.layers.merge.Add object at 0x000001BCDE5206A0> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCDE355D30> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCDE5368D0> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDE5B2828> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCDE5BD390> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCDE5CAEB8> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDE640CC0> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCDE651438> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCDE66D860> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDE6E6128> False
<tensorflow.python.keras.layers.merge.Add object at 0x000001BCDE6E6780> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCDE536F60> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCDE723160> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDE77A940> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCDE787128> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCDE794748> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDE80BDA0> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCDE81E588> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCDE825C18> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDE8B6240> False
<tensorflow.python.keras.layers.merge.Add object at 0x000001BCDE8B6898> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCDE703F98> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCDE8F13C8> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDE945A58> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCDE957240> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCDE960898> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDE9D8EB8> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCDE9EB550> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCDE9F1D30> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDEA81358> False
<tensorflow.python.keras.layers.merge.Add object at 0x000001BCDEA81A20> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCDE8D1BE0> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCDEABCFD0> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDEB15B70> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCDEB23358> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCDEB2B978> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDEBA7FD0> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCDEBBD7B8> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCDEBC1E10> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDEC50438> False
<tensorflow.python.keras.layers.merge.Add object at 0x000001BCDEC50B70> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCDEA9BD30> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCDEC86B38> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDECE2C88> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCDECEE470> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCDECF5AC8> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDFD58128> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCDFD58780> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCDFD5CEF0> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDFDE8588> False
<tensorflow.python.keras.layers.merge.Add object at 0x000001BCDFDF5518> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCDEC6AB38> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCDFE28DA0> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDFE7FDA0> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCDFE8D588> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCDFE94BE0> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDFF24240> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCDFF24898> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCDFF2AC18> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCDFFB7668> False
<tensorflow.python.keras.layers.merge.Add object at 0x000001BCDFFC4D30> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCE005D278> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCE005D978> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCE00E0710> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCE00EF3C8> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCE00FDE80> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCE0173B70> False
<tensorflow.python.keras.layers.convolutional.Conv2D object at 0x000001BCDFFD6550> False
<tensorflow.python.keras.layers.pooling.MaxPooling2D object at 0x000001BCE0186358> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCE004D9B0> False
<tensorflow.python.keras.layers.merge.Add object at 0x000001BCE01A57F0> False
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCE01D6518> False
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCE021CE48> False
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCE0227780> True
<tensorflow.python.keras.layers.convolutional.SeparableConv2D object at 0x000001BCE0230E10> True
<tensorflow.python.keras.layers.normalization.BatchNormalization object at 0x000001BCE02BDEF0> True
<tensorflow.python.keras.layers.core.Activation object at 0x000001BCE02D2588> True
5.7
Epoch 1/30
答案 0 :(得分：0)
由于生成器显示找到的0张图像，因此似乎已磨损发生器。您还应该为验证和测试生成器设置shuffle = False，否则标签的顺序将与图像不匹配。
训练和验证的步骤数应为整数，使用int()进行更改或向上/向下取整。