How to generate tfmasks?

[arunpatro]

Hi,

While generating the tfrecords for the dataset, the code here: https://github.com/zalandoresearch/disentangling_conditional_gans/blob/03d5101ba63ed223d310e42584545bb1d53077e0/dataset.py#L79

loads *.tfmasks. But the code in the repo doesn't create tfmasks. Can someone share the format for tfmasks? Is it a binary valued image for each image? This can be created using segmentation but how to create a tfrecord/tfmask format of it?

Thank you.

[gokyildirim]

I will create an updated branch soon with an example toy dataset, I hope the dataset creation will be clearer.

[gokyildirim]

@arunpatro could you check the new branch "toydata", I hope it is clearer now.

[arunpatro]

Yes, the code is much easier to understand. Thank you so much. However I see that train.py is different than master branch.

On Fri, Jan 25, 2019, 6:44 PM Gokhan Yildirim [email protected] wrote:

@arunpatro https://github.com/arunpatro could you check the new branch "toydata", I hope it is clearer now.

— You are receiving this because you were mentioned.

Reply to this email directly, view it on GitHub https://github.com/zalandoresearch/disentangling_conditional_gans/issues/2#issuecomment-457568966, or mute the thread https://github.com/notifications/unsubscribe-auth/AIBI098iEbu68AixO48EK_xODgA4ZxISks5vGwMrgaJpZM4YbdZG .

[gokyildirim]

yes, I've changed a few things on how the average color is handled. But both branches should lead to compatible trained models.