How is kernel size in keras convolution layers defined?

2
votes

According to the documentation website (https://keras.io/layers/convolutional/) the kernel size of a keras convolution layer is defined as height x width:

kernel_size: An integer or tuple/list of 2 integers, specifying the height and width of the 2D convolution window. Can be a single integer to specify the same value for all spatial dimensions.

In the code documentation however it is defined as the exact opposite:

kernel_size: An integer or tuple/list of 2 integers, specifying the width and height of the 2D convolution window. Can be a single integer to specify the same value for all spatial dimensions.

Which one is true? As far as I figured out, the definition of the website should be the right one, am i right?

1
pythonkerasneural-networkconv-neural-network
What do you mean by code documentation? And I had checked in github arguments for Conv2D. It has the exact information which has in website. - Krunal V
hm interesting. I just copied the description from the installed keras package I am using. Must have been updated.. Thanks! - kleka

1 Answers

1
votes

Yes, the definition in the website is correct (height x width). The code documentation was fixed in version 2.2.0 as per this commit.

https://github.com/keras-team/keras/commit/027d37961afff8a7963836ca87ccdbbf65a78811#diff-d462a3275abee3aa77ba2b341e45cf20L366