by titu1994

titu1994 /keras-coordconv

Keras implementation of CoordConv for all Convolution layers

132 Stars 28 Forks Last release: Not found MIT License 24 Commits 0 Releases

Available items

No Items, yet!

The developer of this repository has not created any items for sale yet. Need a bug fixed? Help with integration? A different license? Create a request here:

CoordConv for Keras

Keras implementation of CoordConv from the paper An intriguing failing of convolutional neural networks and the CoordConv solution.

Extends the

concatenation from only 2D rank (images) to 1D (text / time series) and 3D tensors (video / voxels).


and call it before any convolution layer in order to attach the coordinate channels to the input.

There are 3 different versions of CoordinateChannel - 1D, 2D and 3D for each of

from coord import CoordinateChannel2D

prior to first conv

ip = Input(shape=(64, 64, 2)) x = CoordinateChannel2D()(ip) x = Conv2D(...)(x) # This defines the CoordConv from the paper. ... x = CoordinateChannel2D(use_radius=True)(x) x = Conv2D(...)(x) # This adds the 3rd channel for the radius.


The experiments folder contains the

of a 64x64 grid using the coordinate index as input as in the paper for both

Creating the datasets

First, edit the
file to change the
parameter - to either
. This will generate 2 folders for the datasets and several numpy files.

Uniform Dataset

The uniform dataset model can be trained and evaluated in less than 10 epochs using

|Train | Test | Predictions | |:---: | :---: | :-----------: | | | | |

Quadrant Dataset

The quadrant dataset model can be trained and evaluated in less than 25 epochs using

|Train | Test | Predictions | |:---: | :---: | :-----------: | | | | |


To see if the implementation of CoordConv index concatenation is correct, please refer to the numpy implementations in the

directory, for the implementation of all 3 versions.

Difference from paper

This implementation of the coordinate channels creation differs slightly from the original paper.

The major difference is that for 2/3D Convolutions, it may not be the case that the height and width are the same for all layers. The original implementation would throw an error due to shape mismatch during the concatenation.

To over come this, the

operation which occurs at the first of every channel is modified and a few transpose operations are added to account for this change.

This modification along with some transpose operations allows for height and width to be different and still work.

Theano Support

Theano is partially supported with the
script and using passing a static batch size to the Input layer.


  • Keras 2.2.0+
  • Either Tensorflow or CNTK backend.
  • Matplotlib (to plot images only)

We use cookies. If you continue to browse the site, you agree to the use of cookies. For more information on our use of cookies please see our Privacy Policy.