Skip to content

Commit 202ecfd

Browse files
WZMIAOMIAOdatumbox
andauthored
update paper link for FCOS refence (#6035)
* update paper link for FCOS refence * remove 'updated version' Co-authored-by: Vasilis Vryniotis <[email protected]>
1 parent b430ba6 commit 202ecfd

File tree

1 file changed

+5
-1
lines changed
  • torchvision/models/detection

1 file changed

+5
-1
lines changed

torchvision/models/detection/fcos.py

Lines changed: 5 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -201,7 +201,10 @@ def forward(self, x: List[Tensor]) -> Tensor:
201201

202202
class FCOSRegressionHead(nn.Module):
203203
"""
204-
A regression head for use in FCOS.
204+
A regression head for use in FCOS, which combines regression branch and center-ness branch.
205+
This can obtain better performance.
206+
207+
Reference: `FCOS: A simple and strong anchor-free object detector <https://arxiv.org/abs/2006.09214>`_.
205208
206209
Args:
207210
in_channels (int): number of channels of the input feature
@@ -680,6 +683,7 @@ def fcos_resnet50_fpn(
680683
Constructs a FCOS model with a ResNet-50-FPN backbone.
681684
682685
Reference: `FCOS: Fully Convolutional One-Stage Object Detection <https://arxiv.org/abs/1904.01355>`_.
686+
`FCOS: A simple and strong anchor-free object detector <https://arxiv.org/abs/2006.09214>`_.
683687
684688
The input to the model is expected to be a list of tensors, each of shape ``[C, H, W]``, one for each
685689
image, and should be in ``0-1`` range. Different images can have different sizes.

0 commit comments

Comments
 (0)