expected conv2d to have 4 dimensions, but got array with shape

Question

I am trying to perform a convolutional network on some medical images that have nifti format, using Keras. When I try to fit the model like this:

model.fit(X_train, Y_train, 
      batch_size=batch_size, 
      epochs = n_epoch, 
      validation_data=(X_test, Y_test))

I get this error:

expected conv2d_171 to have 4 dimensions, but got array with shape
(1240, 240, 240)

But when I change the input size from img_channels = 4 to this:

img_channels = 3
img_rows = 240
img_cols = 240

I get another error:

expected input_8 to have shape (240, 240, 3) but got array with shape (240, 240, 4)

The size of the images is like this:

should I resize the images? or flip the order of image elements?

This is the code for the model:

    inputs = Input((img_rows, img_cols, img_channels))
    s = Lambda(lambda x: x / 255) (inputs)

    c1 = Conv2D(16, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (s)
    c1 = Dropout(0.1) (c1)
    c1 = Conv2D(16, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (c1)
    p1 = MaxPooling2D((2, 2)) (c1)

    c2 = Conv2D(32, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (p1)
    c2 = Dropout(0.1) (c2)
    c2 = Conv2D(32, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (c2)
    p2 = MaxPooling2D((2, 2)) (c2)

    c3 = Conv2D(64, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (p2)
    c3 = Dropout(0.2) (c3)
    c3 = Conv2D(64, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (c3)
    p3 = MaxPooling2D((2, 2)) (c3)

    c4 = Conv2D(128, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (p3)
    c4 = Dropout(0.2) (c4)
    c4 = Conv2D(128, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (c4)
    p4 = MaxPooling2D(pool_size=(2, 2)) (c4)

    c5 = Conv2D(256, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (p4)
    c5 = Dropout(0.3) (c5)
    c5 = Conv2D(256, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (c5)

    u6 = Conv2DTranspose(128, (2, 2), strides=(2, 2), padding='same') (c5)
    u6 = concatenate([u6, c4])
    c6 = Conv2D(128, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (u6)
    c6 = Dropout(0.2) (c6)
    c6 = Conv2D(128, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (c6)

    u7 = Conv2DTranspose(64, (2, 2), strides=(2, 2), padding='same') (c6)
    u7 = concatenate([u7, c3])
    c7 = Conv2D(64, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (u7)
    c7 = Dropout(0.2) (c7)
    c7 = Conv2D(64, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (c7)

    u8 = Conv2DTranspose(32, (2, 2), strides=(2, 2), padding='same') (c7)
    u8 = concatenate([u8, c2])
    c8 = Conv2D(32, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (u8)
    c8 = Dropout(0.1) (c8)
    c8 = Conv2D(32, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (c8)

    u9 = Conv2DTranspose(16, (2, 2), strides=(2, 2), padding='same') (c8)
    u9 = concatenate([u9, c1], axis=3)
    c9 = Conv2D(16, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (u9)
    c9 = Dropout(0.1) (c9)
    c9 = Conv2D(16, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (c9)

    outputs = Conv2D(1, (1, 1), activation='sigmoid') (c9)

Asked By: Hadi GhahremanNezhad

||

Source

Answer 1

I see you are using a Unet for segmentation. The problem is with your labels, each of which should be a 3D matrice like (width, height, num_classes). So with the sample length, it should be (sample_length, width, height, num_classes).

In your case, transform them as (310, 240, 240, num_classes). If the label/target images are binary, num_classes=1. Otherwise, you might need to do some preprocessing on the annotated images and one-hot encode them into num_classes.

Also, have a look if you are building up the computation graph multiple times because conv2d_171 seems to be quite a high number. I don’t see that many layers in your architecture.

Answered By: Anakin

expected conv2d to have 4 dimensions, but got array with shape

Question:

Answers: