Skip to content

use generalised_box_iou function to calculate giou_loss #5877

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

Closed
wants to merge 1 commit into from
Closed

use generalised_box_iou function to calculate giou_loss #5877

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
31 changes: 8 additions & 23 deletions torchvision/ops/giou_loss.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,6 +2,7 @@
from torch import Tensor

from ..utils import _log_api_usage_once
from .boxes import generalized_box_iou


def _upcast(t: Tensor) -> Tensor:
Expand Down Expand Up @@ -48,30 +49,14 @@ def generalized_box_iou_loss(

boxes1 = _upcast(boxes1)
boxes2 = _upcast(boxes2)
x1, y1, x2, y2 = boxes1.unbind(dim=-1)
x1g, y1g, x2g, y2g = boxes2.unbind(dim=-1)

# Intersection keypoints
xkis1 = torch.max(x1, x1g)
ykis1 = torch.max(y1, y1g)
xkis2 = torch.min(x2, x2g)
ykis2 = torch.min(y2, y2g)

intsctk = torch.zeros_like(x1)
mask = (ykis2 > ykis1) & (xkis2 > xkis1)
intsctk[mask] = (xkis2[mask] - xkis1[mask]) * (ykis2[mask] - ykis1[mask])
unionk = (x2 - x1) * (y2 - y1) + (x2g - x1g) * (y2g - y1g) - intsctk
iouk = intsctk / (unionk + eps)

# smallest enclosing box
xc1 = torch.min(x1, x1g)
yc1 = torch.min(y1, y1g)
xc2 = torch.max(x2, x2g)
yc2 = torch.max(y2, y2g)

area_c = (xc2 - xc1) * (yc2 - yc1)
miouk = iouk - ((area_c - unionk) / (area_c + eps))

# here we can use diagonal matrix with no problem because the resultant tensor of `generalized_box_iou` will be having shape = N X N
if boxes1.shape == torch.Size([4]):
boxes1 = boxes1[None, :]
boxes2 = boxes2[None, :]
miouk = generalized_box_iou(boxes1, boxes2)[0][0]
else:
miouk = generalized_box_iou(boxes1, boxes2).diagonal()
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is going to be quite slow as noted at #5786 (comment)

Let's continue our discussion on the other thread just to avoid splitting the comments on various places.

loss = 1 - miouk

if reduction == "mean":
Expand Down