Pytorch version of this code and data&annotation(scribble) for the other three datasets?

[JieLiu-UvA]

Hi Gvalvano:

Thanks for your solid and interesting work. I am very interested in your paper, and I wonder whether it is possible for you to provide the Pytorch version of this code as well as data&annotations for the other three datasets?

Best,
Jie

[gvalvano]

Hi Jie :) I'm very happy you like the paper. Answers below:

# PyTorch implementation
Unfortunately, I don't plan to release a PyTorch version of this, for now, but the implementation should be really straightforward! You can see how the attention gates are implemented at every level of the segmentor decoder at this line. Once you obtain the list of attention maps, you can pass them to the discriminator and just concatenate them level by level to the discriminator convolutional blocks, as I do at this line of code.
I hope this will be useful to easily navigate through my code (more info are available in the README file, or you can just ask me ;)).

# Scribbles for the other datasets
Unfortunately, I'm not sure I can release data derived from other datasets before discussing this with the people who have released them! (for the ACDC dataset, we have made sure we could after contacting the ACDC team). In any case, you can download the datasets on your own, and then generate the synthetic scribbles using this code for skeleton-based scribbles, and this code for random walk-based scribbles. With little effort, it should be pretty easy :) But let me know if you need any help

[The-kid-who-loves-to-learn]

Hi Gvalvano:
After reading your article, I would like to know your version of Python? "Because I tried many times and couldn't succeed, thank you."

[gvalvano]

Hi @The-kid-who-loves-to-learn :) this is the environment I tested:

This code was implemented using TensorFlow 1.14. We tested it on a TITAN Xp GPU, and on a GeForce GTX 1080, using CUDA 8.0, 9.0 and 10.2.