'Conv2D' 从 1 中减去 3 导致的负尺寸大小

Question

I'm using Keras with Tensorflow as backend , here is my code:

import numpy as np
np.random.seed(1373) 
import tensorflow as tf
tf.python.control_flow_ops = tf

import os
from keras.datasets import mnist
from keras.models import Sequential
from keras.layers.core import Dense, Dropout, Activation, Flatten
from keras.layers.convolutional import Convolution2D, MaxPooling2D
from keras.utils import np_utils

batch_size = 128
nb_classes = 10
nb_epoch = 12


img_rows, img_cols = 28, 28

nb_filters = 32

nb_pool = 2

nb_conv = 3


(X_train, y_train), (X_test, y_test) = mnist.load_data()

print(X_train.shape[0])

X_train = X_train.reshape(X_train.shape[0], 1, img_rows, img_cols)
X_test = X_test.reshape(X_test.shape[0], 1, img_rows, img_cols)


X_train = X_train.astype('float32')
X_test = X_test.astype('float32')
X_train /= 255
X_test /= 255


print('X_train shape:', X_train.shape)
print(X_train.shape[0], 'train samples')
print(X_test.shape[0], 'test samples')


Y_train = np_utils.to_categorical(y_train, nb_classes)
Y_test = np_utils.to_categorical(y_test, nb_classes)

model = Sequential()

model.add(Convolution2D(nb_filters, nb_conv, nb_conv,
border_mode='valid',
input_shape=(1, img_rows, img_cols)))
model.add(Activation('relu'))
model.add(Convolution2D(nb_filters, nb_conv, nb_conv))
model.add(Activation('relu'))

model.add(MaxPooling2D(pool_size=(nb_pool, nb_pool)))
model.add(Dropout(0.25))

model.add(Flatten())
model.add(Dense(128))
model.add(Activation('relu'))
model.add(Dropout(0.5))
model.add(Dense(nb_classes)) 
model.add(Activation('softmax')) 

model.compile(loss='categorical_crossentropy', optimizer='adadelta', metrics=["accuracy"])


model.fit(X_train, Y_train, batch_size=batch_size, nb_epoch=nb_epoch,
verbose=1, validation_data=(X_test, Y_test))

score = model.evaluate(X_test, Y_test, verbose=0)

print('Test score:', score[0])
print('Test accuracy:', score[1])

and Trackback error:

Using TensorFlow backend.
60000
('X_train shape:', (60000, 1, 28, 28))
(60000, 'train samples')
(10000, 'test samples')
Traceback (most recent call last):
  File "mnist.py", line 154, in <module>
    input_shape=(1, img_rows, img_cols)))
  File "/usr/local/lib/python2.7/dist-packages/keras/models.py", line 276, in add
    layer.create_input_layer(batch_input_shape, input_dtype)
  File "/usr/local/lib/python2.7/dist-packages/keras/engine/topology.py", line 370, in create_input_layer
    self(x)
  File "/usr/local/lib/python2.7/dist-packages/keras/engine/topology.py", line 514, in __call__
    self.add_inbound_node(inbound_layers, node_indices, tensor_indices)
  File "/usr/local/lib/python2.7/dist-packages/keras/engine/topology.py", line 572, in add_inbound_node
    Node.create_node(self, inbound_layers, node_indices, tensor_indices)
  File "/usr/local/lib/python2.7/dist-packages/keras/engine/topology.py", line 149, in create_node
    output_tensors = to_list(outbound_layer.call(input_tensors[0], mask=input_masks[0]))
  File "/usr/local/lib/python2.7/dist-packages/keras/layers/convolutional.py", line 466, in call
    filter_shape=self.W_shape)
  File "/usr/local/lib/python2.7/dist-packages/keras/backend/tensorflow_backend.py", line 1579, in conv2d
    x = tf.nn.conv2d(x, kernel, strides, padding=padding)
  File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/ops/gen_nn_ops.py", line 396, in conv2d
    data_format=data_format, name=name)
  File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/framework/op_def_library.py", line 759, in apply_op
    op_def=op_def)
  File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/framework/ops.py", line 2242, in create_op
    set_shapes_for_outputs(ret)
  File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/framework/ops.py", line 1617, in set_shapes_for_outputs
    shapes = shape_func(op)
  File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/framework/ops.py", line 1568, in call_with_requiring
    return call_cpp_shape_fn(op, require_shape_fn=True)
  File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/framework/common_shapes.py", line 610, in call_cpp_shape_fn
    debug_python_shape_fn, require_shape_fn)
  File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/framework/common_shapes.py", line 675, in _call_cpp_shape_fn_impl
    raise ValueError(err.message)
ValueError: Negative dimension size caused by subtracting 3 from 1 for 'Conv2D' (op: 'Conv2D') with input shapes: [?,1,28,28], [3,3,28,32].

First I saw some answers that problem is with Tensorflow version so I upgrade Tensorflow to 0.12.0 , but still exist , is that problem with network or I missing something, what should input_shape looks like?

Update Here is ./keras/keras.json :

{
    "image_dim_ordering": "tf", 
    "epsilon": 1e-07, 
    "floatx": "float32", 
    "backend": "tensorflow"
}

Answer 1

Your issue comes from the image_ordering_dim in keras.json .

From Keras Image Processing doc :

dim_ordering: One of {"th", "tf"}. "tf" mode means that the images should have shape (samples, height, width, channels), "th" mode means that the images should have shape (samples, channels, height, width). It defaults to the image_dim_ordering value found in your Keras config file at ~/.keras/keras.json. If you never set it, then it will be "tf".

Keras maps the convolution operation to the chosen backend (theano or tensorflow). However, both backends have made different choices for the ordering of the dimensions. If your image batch is of N images of HxW size with C channels, theano uses the NCHW ordering while tensorflow uses the NHWC ordering.

Keras allows you to choose which ordering you prefer and will do the conversion to map to the backends behind. But if you choose image_ordering_dim="th" it expects Theano-style ordering (NCHW, the one you have in your code) and if image_ordering_dim="tf" it expects tensorflow-style ordering (NHWC).

Since your image_ordering_dim is set to "tf" , if you reshape your data to the tensorflow style it should work:

X_train = X_train.reshape(X_train.shape[0], img_cols, img_rows, 1)
X_test = X_test.reshape(X_test.shape[0], img_cols, img_rows, 1)

and

input_shape=(img_cols, img_rows, 1)

Answer 2

FWIW，我用一些 strides 或 kernel_size 但不是全部的值反复得到这个错误，后端和 image_ordering 已经设置为张量流，当我添加padding="same"时它们都消失了

Answer 3

Just add this:

from keras import backend as K
K.set_image_dim_ordering('th')

Answer 4

I am also having the same problem. However, each Conv3D layer, I am using, is reducing size of the input. So, including one parameter padding='same' during declaring the Conv2D/3D layer solved the problem. Here is the demo code

model.add(Conv3D(32,kernel_size=(3,3,3),activation='relu',padding='same'))

Reducing the size of the filter can also solve the problem.

Actually, Conv3D or Conv2D layer reduces the input data. But when your next layer does not recieve any input or input of size which is not appropriate for that layer, then this error occurs. By padding we are making the output of Conv3Dor2D remain same size of input so that next layer will get its desired input

Answer 5

I faced the same problem, but it was solved by changing the conv2d function:

if K.image_data_format=='channels_first':
    x_train = x_train.reshape(x_train.shape[0], 1,img_cols,img_rows)
    x_test = x_test.reshape(x_test.shape[0], 1,img_cols,img_rows)
    input_shape = (1,img_cols,img_rows)
else:
    x_train = x_train.reshape(x_train.shape[0],img_cols,img_rows,1)
    x_test = x_test.reshape(x_test.shape[0],img_cols,img_rows,1)
    input_shape = (img_cols,img_rows,1)
model.add(Convolution2D(32,(3, 3), input_shape = input_shape, activation="relu"))

Answer 6

Provide size of filter using parenthesis like:

model.add(Convolution2D(nb_filters,( nb_conv, nb_conv) ,border_mode='valid',
input_shape=(1, img_rows, img_cols)))

It will work in my case and also change the X_train , X_test as this:

X_train = X_train.reshape(X_train.shape[0], img_cols, img_rows, 1)
X_test = X_test.reshape(X_test.shape[0], img_cols, img_rows, 1)

Answer 7

Another solution can help is change:

from keras.layers import Convolution2D, MaxPooling2D

to

from keras.layers import Conv2D, MaxPooling2D

After that, to run preprocess input data, I change:

X_train = X_train.reshape(X_train.shape[0], 1, 28, 28)
X_test = X_test.reshape(X_test.shape[0], 1, 28, 28)

to

X_train = X_train.reshape(X_train.shape[0], 28, 28, 1)
X_test.reshape(X_test.shape[0], 28, 28, 1)

Finally, I change:

model.add(Convolution2D(32, 3, 3, activation='relu',input_shape=(1,28,28))) 
model.add(Convolution2D(32, 3, 3,activation='relu'))

to

model.add(Conv2D(32, (3, 3), activation='relu',input_shape=(28,28,1)))
model.add(Conv2D(32, (3, 3), activation='relu'))

Answer 8

Just had this problem. The bellow is the solution to work with the new API.

K.set_image_dim_ordering('tf') --> K.set_image_data_format('channels_last')
K.set_image_dim_ordering('th') --> K.set_image_data_format('channels_first')
K.image_dim_ordering() == 'tf' --> K.image_data_format() == 'channels_last'
K.image_dim_ordering() == 'th' --> K.image_data_format() == 'channels_first'

See more here

Answer 9

   %store -r le
   %store -r x_train 
   %store -r x_test 
   %store -r y_train 
   %store -r y_test 
   %store -r yy 
    import numpy as np
    import keras
    from keras.models import Sequential
    from keras.layers import Dense, Dropout, Activation, Flatten
    from keras.layers import Conv2D, MaxPooling2D, GlobalAveragePooling2D
    from keras.optimizers import Adam
    from keras.utils import np_utils
    from sklearn import metrics
    num_rows = 40
    num_columns = 174
    num_channels = 1
    x_train = x_train.reshape(x_train.shape[0],num_rows , num_columns, num_channels)
    x_test = x_test.reshape(x_test.shape[0], num_rows, num_columns,num_channels )
num_labels = yy.shape[1]
filter_size = 2
# Construct model 
model = Sequential()

model.add(Conv2D(filters=16, kernel_size=2, activation='relu',input_shape=( 
40,174,1)))
model.add(MaxPooling2D(pool_size=2))
model.add(Dropout(0.2))

model.add(Conv2D(filters=32, kernel_size=2, activation='relu'))
model.add(MaxPooling2D(pool_size=2))
model.add(Dropout(0.2))

model.add(Conv2D(filters=64, kernel_size=2, activation='relu'))
model.add(MaxPooling2D(pool_size=2))
model.add(Dropout(0.2))

model.add(Conv2D(filters=128, kernel_size=2, activation='relu'))
model.add(MaxPooling2D(pool_size=2))
model.add(Dropout(0.2))
model.add(GlobalAveragePooling2D())

model.add(Dense(num_labels, activation='softmax'))