U-2-Net model different with your paper description

[laoyoutiaotiao]

https://github.com/NathanUA/U-2-Net/blob/master/model/u2net.py

    d1 = self.side1(hx1d)
    
    d2 = self.side2(hx2d)
    d2 = _upsample_like(d2,d1)

    d3 = self.side3(hx3d)
    d3 = _upsample_like(d3,d1)

    d4 = self.side4(hx4d)
    d4 = _upsample_like(d4,d1)

    d5 = self.side5(hx5d)
    d5 = _upsample_like(d5,d1)

    d6 = self.side6(hx6)
    d6 = _upsample_like(d6,d1)

    d0 = self.outconv(torch.cat((d1,d2,d3,d4,d5,d6),1))

    return F.sigmoid(d0), F.sigmoid(d1), F.sigmoid(d2), F.sigmoid(d3), F.sigmoid(d4), F.sigmoid(d5), F.sigmoid(d6)

generates six side output saliency probability maps from stages En6, De5, De4, De3, De2 and De1 by a 3x3 convolution layer, without sigmoid function.
But in your paper description，generates six side output saliency probability maps from stages En6, De5, De4, De3, De2 and De1 by a 3x3 convolution layer and a sigmoid function.

Why the difference?

[bluesky314]

There seems to be sigmoids here:
" return F.sigmoid(d0), F.sigmoid(d1), F.sigmoid(d2), F.sigmoid(d3), F.sigmoid(d4), F.sigmoid(d5), F.sigmoid(d6)"

[laoyoutiaotiao]

in the last convolution layer 'd0 = self.outconv(torch.cat((d1,d2,d3,d4,d5,d6),1))', d1d6 just from stages En6De1 by a 3x3 convolution layer , without sigmoid function. Is that understanding？

[xuebinqin]

Thanks for you insightful comments. Yes, here the code is a bit different from the description in the paper. Because drawing both logits and probability maps takes more space in the figure of the paper. For the simplicity, we just draw the probability maps (of d1-d6) there and then gave inaccurate descriptions based on the figure. We are sorry about that. Please follow the released code for usage.

…

On Aug 4, 2020, at 3:33 AM, laoyoutiaotiao ***@***.***> wrote: https://github.com/NathanUA/U-2-Net/blob/master/model/u2net.py <x-msg://2/url> ` #side output d1 = self.side1(hx1d) d2 = self.side2(hx2d) d2 = _upsample_like(d2,d1) d3 = self.side3(hx3d) d3 = _upsample_like(d3,d1) d4 = self.side4(hx4d) d4 = _upsample_like(d4,d1) d5 = self.side5(hx5d) d5 = _upsample_like(d5,d1) d6 = self.side6(hx6) d6 = _upsample_like(d6,d1) d0 = self.outconv(torch.cat((d1,d2,d3,d4,d5,d6),1)) return F.sigmoid(d0), F.sigmoid(d1), F.sigmoid(d2), F.sigmoid(d3), F.sigmoid(d4), F.sigmoid(d5), F.sigmoid(d6)` generates six side output saliency probability maps from stages En6, De5, De4, De3, De2 and De1 by a 3x3 convolution layer, without sigmoid function. But in your paper description，generates six side output saliency probability maps from stages En6, De5, De4, De3, De2 and De1 by a 3x3 convolution layer and a sigmoid function. Why the difference? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#52>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADSGORPIT64JELI5D7GCQY3R67IW3ANCNFSM4PUGQLQQ>.