Skip to content

Added variable output dim to the conditioned channel attention module #11

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
rymuelle merged 1 commit into from
Nov 11, 2025
Merged

Added variable output dim to the conditioned channel attention module #11

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 6 additions & 4 deletions src/Restorer/Cond_NAF.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -50,10 +50,12 @@ def forward(self, x):


class ConditionedChannelAttention(nn.Module):
def __init__(self, dims, cat_dims):
def __init__(self, dims, cat_dims, out_dim=0):
super().__init__()
in_dim = dims + cat_dims
self.mlp = nn.Sequential(nn.Linear(in_dim, dims))
if not out_dim:
out_dim = dims
self.mlp = nn.Sequential(nn.Linear(in_dim, out_dim))
self.pool = nn.AdaptiveAvgPool2d(1)

def forward(self, x, conditioning):
Expand Down Expand Up @@ -95,7 +97,7 @@ def forward(self, x):
class CondFuser(nn.Module):
def __init__(self, chan, cond_chan=1):
super().__init__()
self.cca = ConditionedChannelAttention(chan * 2, cond_chan)
self.cca = ConditionedChannelAttention(chan * 2, cond_chan, out_dim=chan)
self.sig = nn.Sigmoid()

self.sa = nn.Sequential(
Expand All @@ -107,7 +109,7 @@ def __init__(self, chan, cond_chan=1):

def forward(self, x1, x2, cond):
x = torch.cat([x1, x2], dim=1)
x2 = 1 * self.sig(self.cca(x)) * self.sa(x) * x2
x2 = 1 * self.sig(self.cca(x, cond)) * self.sa(x) * x2
return x1 + x2


Expand Down
Loading