Fashion-MNIST 数据集维度的困难

Question

does someone know how to fix this error? I tried to solve it with the reshape function but it still doesnt work. I want to train the Fashion-MNIST dataset with VGG16 but I have some difficulties with the dimensions...

I get this error:

Epoch 1/50

---------------------------------------------------------------------------

ValueError                                Traceback (most recent call last)

<ipython-input-84-51241dcc88e5> in <module>()
      3                       steps_per_epoch=x_train.shape[0]//batch_size,
      4                       validation_data=val_generator.flow(x_val,y_val,batch_size=batch_size),validation_steps=250,
----> 5                       callbacks=[lrr],verbose=1)

5 frames

/usr/local/lib/python3.6/dist-packages/keras/engine/training_utils.py in standardize_input_data(data, names, shapes, check_batch_axis, exception_prefix)
    143                             ': expected ' + names[i] + ' to have shape ' +
    144                             str(shape) + ' but got array with shape ' +
--> 145                             str(data_shape))
    146     return data
    147 

ValueError: Error when checking input: expected vgg16_input to have shape (32, 32, 3) but got array with shape (28, 28, 1)

And when I try to fix it with the reshape function I get this error:

ValueError                                Traceback (most recent call last)

<ipython-input-11-8ad7b8e95a75> in <module>()
      1 #x_train = x_train.reshape(x_train.shape[0],32,32,3)
----> 2 x_train = x_train.reshape(-1,32,32,3)

ValueError: cannot reshape array of size 32928000 into shape (32,32,3)

My input has the shape:

(42000, 28, 28, 1)

Answer 1

ValueError: Error when checking input: expected vgg16_input to have shape (32, 32, 3) but got array with shape (28, 28, 1)

VGG is expecting the input to be a 32*32 patch;

ValueError: cannot reshape array of size 32928000 into shape (32,32,3)

This happens because the array shape isn't completely divisible by the product of your 32*32*3 to get the 4th axis;

My input has the shape: (42000, 28, 28, 1)

Then change (if possible) the vgg_input to accept (28,28,3) patches etc; (accordingly)

Sample Code from here ,

def build(width, height, depth, classes):
    # initialize the model along with the input shape to be
    # "channels last" and the channels dimension itself
    model = Sequential()
    inputShape = (height, width, depth)
    chanDim = -1
    # if we are using "channels first", update the input shape
    # and channels dimension
    if K.image_data_format() == "channels_first":
        inputShape = (depth, height, width)
        chanDim = 1

This article is also a good read!