Very thanks for sharing your code. But when I download the Pre-computed results, each map is the background map. Could you upload the Pre-computed results again? Thank you~

I encounter "You don't have permission to access this resource." when I visit the pretrained weights' website.
Could you please help me get the pretrained model? Thanks.

Thanks for your nice work, where is the training code?

Dear author:

could you please share me the training code? Thank you in advance...

hello, after reading your paper, I don't understand how to train the model.
In experiment section, "binary cross-entropy loss" is mentioned, it seems the network should be trained with pixel-wise supervised label like segmentation.
Could you show the "unsupervised process" with detail? Thank you.

Thanks for your contribution, I just want to know when will the code be available?

Hi Zhao Yang & Qiang Wang,

Thank you for your impressive work. I evaluated the model by following your instruction in readme. However the masks (png files) in pred_masks were all black. Why did it get these black masks while the mean J was about 0.8? Thank you!

Just more similarity matrices, more memory? Do the concatenation approach instead of the multiplication?

I use pretrained model to inference: python eval.py --ms-mirror True --inst-prune True --eval-sal True

and use matlab code to evalate, the result is as follows:
cosnet adnet mmadnet

J mean 0.805 0.817 0.803
J recall 0.931 0.909 0.900
J decay 0.044 0.022 0.021

F mean 0.795 0.805 0.793
F recall 0.895 0.851 0.847
F decay 0.050 0.006 0.004

T (GT 0.088) 0.184 0.225 0.228

adnet is from your provided files. mmadnet is from me.
I am not sure what is wrong. Is it some parameters wrong?

hello , can you push the training code to my E-mail, thanks!

[email protected]

Hi, I am confused about how to optimize.
1, what's the loss function? in this paper, it seems not to be mentioned. is it just BCE?
2, what's the ground truth?
Thank you!