HongChuang/ultralytics-ascend
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 3

Update IoU capitalization (#8604)

Browse source 
Signed-off-by: Glenn Jocher <glenn.jocher@ultralytics.com>
Co-authored-by: Dean Mark <2552482+deanmark@users.noreply.github.com>
This commit is contained in:
Glenn Jocher 2024-03-02 22:12:23 +01:00 • committed by GitHub
parent e0b8b36967
commit 1146bb0582
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
8 changed files with 16 additions and 16 deletions
Download patch file Download diff file Expand all files Collapse all files

6
ultralytics/utils/metrics.py
View file

@ -24,7 +24,7 @@ def bbox_ioa(box1, box2, iou=False, eps=1e-7):
Args:
box1 (np.ndarray): A numpy array of shape (n, 4) representing n bounding boxes.
box2 (np.ndarray): A numpy array of shape (m, 4) representing m bounding boxes.
iou (bool): Calculate the standard iou if True else return inter_area/box2_area.
iou (bool): Calculate the standard IoU if True else return inter_area/box2_area.
eps (float, optional): A small value to avoid division by zero. Defaults to 1e-7.
Returns:
@ -194,7 +194,7 @@ def _get_covariance_matrix(boxes):
def probiou(obb1, obb2, CIoU=False, eps=1e-7):
"""
Calculate the prob iou between oriented bounding boxes, https://arxiv.org/pdf/2106.06072v1.pdf.
Calculate the prob IoU between oriented bounding boxes, https://arxiv.org/pdf/2106.06072v1.pdf.
Args:
obb1 (torch.Tensor): A tensor of shape (N, 5) representing ground truth obbs, with xywhr format.
@ -233,7 +233,7 @@ def probiou(obb1, obb2, CIoU=False, eps=1e-7):
def batch_probiou(obb1, obb2, eps=1e-7):
"""
Calculate the prob iou between oriented bounding boxes, https://arxiv.org/pdf/2106.06072v1.pdf.
Calculate the prob IoU between oriented bounding boxes, https://arxiv.org/pdf/2106.06072v1.pdf.
Args:
obb1 (torch.Tensor | np.ndarray): A tensor of shape (N, 5) representing ground truth obbs, with xywhr format.

4
ultralytics/utils/ops.py
View file

@ -147,7 +147,7 @@ def nms_rotated(boxes, scores, threshold=0.45):
Args:
boxes (torch.Tensor): (N, 5), xywhr.
scores (torch.Tensor): (N, ).
threshold (float): Iou threshold.
threshold (float): IoU threshold.
Returns:
"""
@ -287,7 +287,7 @@ def non_max_suppression(
# if merge and (1 < n < 3E3): # Merge NMS (boxes merged using weighted mean)
# # Update boxes as boxes(i,4) = weights(i,n) * boxes(n,4)
# from .metrics import box_iou
# iou = box_iou(boxes[i], boxes) > iou_thres # iou matrix
# iou = box_iou(boxes[i], boxes) > iou_thres # IoU matrix
# weights = iou * scores[None] # box weights
# x[i, :4] = torch.mm(weights, x[:, :4]).float() / weights.sum(1, keepdim=True) # merged boxes
# redundant = True # require redundant detections

6
ultralytics/utils/tal.py
View file

@ -121,7 +121,7 @@ class TaskAlignedAssigner(nn.Module):
return align_metric, overlaps
def iou_calculation(self, gt_bboxes, pd_bboxes):
"""Iou calculation for horizontal bounding boxes."""
"""IoU calculation for horizontal bounding boxes."""
return bbox_iou(gt_bboxes, pd_bboxes, xywh=False, CIoU=True).squeeze(-1).clamp_(0)
def select_topk_candidates(self, metrics, largest=True, topk_mask=None):
@ -231,7 +231,7 @@ class TaskAlignedAssigner(nn.Module):
@staticmethod
def select_highest_overlaps(mask_pos, overlaps, n_max_boxes):
"""
If an anchor box is assigned to multiple gts, the one with the highest IoI will be selected.
If an anchor box is assigned to multiple gts, the one with the highest IoU will be selected.
Args:
mask_pos (Tensor): shape(b, n_max_boxes, h*w)
@ -260,7 +260,7 @@ class TaskAlignedAssigner(nn.Module):
class RotatedTaskAlignedAssigner(TaskAlignedAssigner):
def iou_calculation(self, gt_bboxes, pd_bboxes):
"""Iou calculation for rotated bounding boxes."""
"""IoU calculation for rotated bounding boxes."""
return probiou(gt_bboxes, pd_bboxes).squeeze(-1).clamp_(0)
@staticmethod