Yao Qin comments

Results 7 comments of


                                            Yao Qin

problem

Hi, our experiments are designed to compare autofocus layer and dilated convolution and the results show that autofocus layer is better than dilated convolution.

problem

Yes, this is what we claimed in our paper. A standard convolutional layer or dilated convolutional layer can be replaced by our autofocus layer to achieve better performance.

test_full.py working but results in exception

Hi Jonathan, Thanks for your contact about the bug. Please just remove the line 112 in test "return score, pred_seg". It should work now. If not, please let me know...

Questions about the setting

I think the default setting is the setting we used.

this paper was accepted in MICCAI 2018, it's worng about citation in your README.

thanks!

is it possible to working with CPU ?

I have not tried it on the CPU. The code is initially designed for GPU. You might need to do some modifications in order to run on CPU and I...

question about the autofocus implementation

Here we do not want the attention map to influence the weights update in the main branch. Therefore, we use x.detach to treat the features extracted from the main branch...