lichen/detecttracking
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

add yolo v10 and modify pipeline

Browse Source
This commit is contained in:
王庆刚
2025-03-28 13:19:54 +08:00
parent 183299c06b
commit 798c596acc
471 changed files with 19109 additions and 7342 deletions
Download Patch File Download Diff File Expand all files Collapse all files

210
ultralytics/utils/tal.py
View File

@ -4,65 +4,18 @@ import torch
import torch.nn as nn
from .checks import check_version
from .metrics import bbox_iou
from .metrics import bbox_iou, probiou
from .ops import xywhr2xyxyxyxy
TORCH_1_10 = check_version(torch.__version__, '1.10.0')
def select_candidates_in_gts(xy_centers, gt_bboxes, eps=1e-9):
"""
Select the positive anchor center in gt.
Args:
xy_centers (Tensor): shape(h*w, 2)
gt_bboxes (Tensor): shape(b, n_boxes, 4)
Returns:
(Tensor): shape(b, n_boxes, h*w)
"""
n_anchors = xy_centers.shape[0]
bs, n_boxes, _ = gt_bboxes.shape
lt, rb = gt_bboxes.view(-1, 1, 4).chunk(2, 2) # left-top, right-bottom
bbox_deltas = torch.cat((xy_centers[None] - lt, rb - xy_centers[None]), dim=2).view(bs, n_boxes, n_anchors, -1)
# return (bbox_deltas.min(3)[0] > eps).to(gt_bboxes.dtype)
return bbox_deltas.amin(3).gt_(eps)
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.
Args:
mask_pos (Tensor): shape(b, n_max_boxes, h*w)
overlaps (Tensor): shape(b, n_max_boxes, h*w)
Returns:
target_gt_idx (Tensor): shape(b, h*w)
fg_mask (Tensor): shape(b, h*w)
mask_pos (Tensor): shape(b, n_max_boxes, h*w)
"""
# (b, n_max_boxes, h*w) -> (b, h*w)
fg_mask = mask_pos.sum(-2)
if fg_mask.max() > 1: # one anchor is assigned to multiple gt_bboxes
mask_multi_gts = (fg_mask.unsqueeze(1) > 1).expand(-1, n_max_boxes, -1) # (b, n_max_boxes, h*w)
max_overlaps_idx = overlaps.argmax(1) # (b, h*w)
is_max_overlaps = torch.zeros(mask_pos.shape, dtype=mask_pos.dtype, device=mask_pos.device)
is_max_overlaps.scatter_(1, max_overlaps_idx.unsqueeze(1), 1)
mask_pos = torch.where(mask_multi_gts, is_max_overlaps, mask_pos).float() # (b, n_max_boxes, h*w)
fg_mask = mask_pos.sum(-2)
# Find each grid serve which gt(index)
target_gt_idx = mask_pos.argmax(-2) # (b, h*w)
return target_gt_idx, fg_mask, mask_pos
TORCH_1_10 = check_version(torch.__version__, "1.10.0")
class TaskAlignedAssigner(nn.Module):
"""
A task-aligned assigner for object detection.
This class assigns ground-truth (gt) objects to anchors based on the task-aligned metric,
which combines both classification and localization information.
This class assigns ground-truth (gt) objects to anchors based on the task-aligned metric, which combines both
classification and localization information.
Attributes:
topk (int): The number of top candidates to consider.
@ -85,8 +38,8 @@ class TaskAlignedAssigner(nn.Module):
@torch.no_grad()
def forward(self, pd_scores, pd_bboxes, anc_points, gt_labels, gt_bboxes, mask_gt):
"""
Compute the task-aligned assignment.
Reference https://github.com/Nioolek/PPYOLOE_pytorch/blob/master/ppyoloe/assigner/tal_assigner.py
Compute the task-aligned assignment. Reference code is available at
https://github.com/Nioolek/PPYOLOE_pytorch/blob/master/ppyoloe/assigner/tal_assigner.py.
Args:
pd_scores (Tensor): shape(bs, num_total_anchors, num_classes)
@ -103,19 +56,24 @@ class TaskAlignedAssigner(nn.Module):
fg_mask (Tensor): shape(bs, num_total_anchors)
target_gt_idx (Tensor): shape(bs, num_total_anchors)
"""
self.bs = pd_scores.size(0)
self.n_max_boxes = gt_bboxes.size(1)
self.bs = pd_scores.shape[0]
self.n_max_boxes = gt_bboxes.shape[1]
if self.n_max_boxes == 0:
device = gt_bboxes.device
return (torch.full_like(pd_scores[..., 0], self.bg_idx).to(device), torch.zeros_like(pd_bboxes).to(device),
torch.zeros_like(pd_scores).to(device), torch.zeros_like(pd_scores[..., 0]).to(device),
torch.zeros_like(pd_scores[..., 0]).to(device))
return (
torch.full_like(pd_scores[..., 0], self.bg_idx).to(device),
torch.zeros_like(pd_bboxes).to(device),
torch.zeros_like(pd_scores).to(device),
torch.zeros_like(pd_scores[..., 0]).to(device),
torch.zeros_like(pd_scores[..., 0]).to(device),
)
mask_pos, align_metric, overlaps = self.get_pos_mask(pd_scores, pd_bboxes, gt_labels, gt_bboxes, anc_points,
mask_gt)
mask_pos, align_metric, overlaps = self.get_pos_mask(
pd_scores, pd_bboxes, gt_labels, gt_bboxes, anc_points, mask_gt
)
target_gt_idx, fg_mask, mask_pos = select_highest_overlaps(mask_pos, overlaps, self.n_max_boxes)
target_gt_idx, fg_mask, mask_pos = self.select_highest_overlaps(mask_pos, overlaps, self.n_max_boxes)
# Assigned target
target_labels, target_bboxes, target_scores = self.get_targets(gt_labels, gt_bboxes, target_gt_idx, fg_mask)
@ -131,7 +89,7 @@ class TaskAlignedAssigner(nn.Module):
def get_pos_mask(self, pd_scores, pd_bboxes, gt_labels, gt_bboxes, anc_points, mask_gt):
"""Get in_gts mask, (b, max_num_obj, h*w)."""
mask_in_gts = select_candidates_in_gts(anc_points, gt_bboxes)
mask_in_gts = self.select_candidates_in_gts(anc_points, gt_bboxes)
# Get anchor_align metric, (b, max_num_obj, h*w)
align_metric, overlaps = self.get_box_metrics(pd_scores, pd_bboxes, gt_labels, gt_bboxes, mask_in_gts * mask_gt)
# Get topk_metric mask, (b, max_num_obj, h*w)
@ -157,11 +115,15 @@ class TaskAlignedAssigner(nn.Module):
# (b, max_num_obj, 1, 4), (b, 1, h*w, 4)
pd_boxes = pd_bboxes.unsqueeze(1).expand(-1, self.n_max_boxes, -1, -1)[mask_gt]
gt_boxes = gt_bboxes.unsqueeze(2).expand(-1, -1, na, -1)[mask_gt]
overlaps[mask_gt] = bbox_iou(gt_boxes, pd_boxes, xywh=False, CIoU=True).squeeze(-1).clamp_(0)
overlaps[mask_gt] = self.iou_calculation(gt_boxes, pd_boxes)
align_metric = bbox_scores.pow(self.alpha) * overlaps.pow(self.beta)
return align_metric, overlaps
def iou_calculation(self, gt_bboxes, pd_bboxes):
"""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):
"""
Select the top-k candidates based on the given metrics.
@ -191,9 +153,9 @@ class TaskAlignedAssigner(nn.Module):
ones = torch.ones_like(topk_idxs[:, :, :1], dtype=torch.int8, device=topk_idxs.device)
for k in range(self.topk):
# Expand topk_idxs for each value of k and add 1 at the specified positions
count_tensor.scatter_add_(-1, topk_idxs[:, :, k:k + 1], ones)
count_tensor.scatter_add_(-1, topk_idxs[:, :, k : k + 1], ones)
# count_tensor.scatter_add_(-1, topk_idxs, torch.ones_like(topk_idxs, dtype=torch.int8, device=topk_idxs.device))
# filter invalid bboxes
# Filter invalid bboxes
count_tensor.masked_fill_(count_tensor > 1, 0)
return count_tensor.to(metrics.dtype)
@ -229,15 +191,17 @@ class TaskAlignedAssigner(nn.Module):
target_labels = gt_labels.long().flatten()[target_gt_idx] # (b, h*w)
# Assigned target boxes, (b, max_num_obj, 4) -> (b, h*w, 4)
target_bboxes = gt_bboxes.view(-1, 4)[target_gt_idx]
target_bboxes = gt_bboxes.view(-1, gt_bboxes.shape[-1])[target_gt_idx]
# Assigned target scores
target_labels.clamp_(0)
# 10x faster than F.one_hot()
target_scores = torch.zeros((target_labels.shape[0], target_labels.shape[1], self.num_classes),
dtype=torch.int64,
device=target_labels.device) # (b, h*w, 80)
target_scores = torch.zeros(
(target_labels.shape[0], target_labels.shape[1], self.num_classes),
dtype=torch.int64,
device=target_labels.device,
) # (b, h*w, 80)
target_scores.scatter_(2, target_labels.unsqueeze(-1), 1)
fg_scores_mask = fg_mask[:, :, None].repeat(1, 1, self.num_classes) # (b, h*w, 80)
@ -245,6 +209,87 @@ class TaskAlignedAssigner(nn.Module):
return target_labels, target_bboxes, target_scores
@staticmethod
def select_candidates_in_gts(xy_centers, gt_bboxes, eps=1e-9):
"""
Select the positive anchor center in gt.
Args:
xy_centers (Tensor): shape(h*w, 2)
gt_bboxes (Tensor): shape(b, n_boxes, 4)
Returns:
(Tensor): shape(b, n_boxes, h*w)
"""
n_anchors = xy_centers.shape[0]
bs, n_boxes, _ = gt_bboxes.shape
lt, rb = gt_bboxes.view(-1, 1, 4).chunk(2, 2) # left-top, right-bottom
bbox_deltas = torch.cat((xy_centers[None] - lt, rb - xy_centers[None]), dim=2).view(bs, n_boxes, n_anchors, -1)
# return (bbox_deltas.min(3)[0] > eps).to(gt_bboxes.dtype)
return bbox_deltas.amin(3).gt_(eps)
@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 IoU will be selected.
Args:
mask_pos (Tensor): shape(b, n_max_boxes, h*w)
overlaps (Tensor): shape(b, n_max_boxes, h*w)
Returns:
target_gt_idx (Tensor): shape(b, h*w)
fg_mask (Tensor): shape(b, h*w)
mask_pos (Tensor): shape(b, n_max_boxes, h*w)
"""
# (b, n_max_boxes, h*w) -> (b, h*w)
fg_mask = mask_pos.sum(-2)
if fg_mask.max() > 1: # one anchor is assigned to multiple gt_bboxes
mask_multi_gts = (fg_mask.unsqueeze(1) > 1).expand(-1, n_max_boxes, -1) # (b, n_max_boxes, h*w)
max_overlaps_idx = overlaps.argmax(1) # (b, h*w)
is_max_overlaps = torch.zeros(mask_pos.shape, dtype=mask_pos.dtype, device=mask_pos.device)
is_max_overlaps.scatter_(1, max_overlaps_idx.unsqueeze(1), 1)
mask_pos = torch.where(mask_multi_gts, is_max_overlaps, mask_pos).float() # (b, n_max_boxes, h*w)
fg_mask = mask_pos.sum(-2)
# Find each grid serve which gt(index)
target_gt_idx = mask_pos.argmax(-2) # (b, h*w)
return target_gt_idx, fg_mask, mask_pos
class RotatedTaskAlignedAssigner(TaskAlignedAssigner):
def iou_calculation(self, gt_bboxes, pd_bboxes):
"""IoU calculation for rotated bounding boxes."""
return probiou(gt_bboxes, pd_bboxes).squeeze(-1).clamp_(0)
@staticmethod
def select_candidates_in_gts(xy_centers, gt_bboxes):
"""
Select the positive anchor center in gt for rotated bounding boxes.
Args:
xy_centers (Tensor): shape(h*w, 2)
gt_bboxes (Tensor): shape(b, n_boxes, 5)
Returns:
(Tensor): shape(b, n_boxes, h*w)
"""
# (b, n_boxes, 5) --> (b, n_boxes, 4, 2)
corners = xywhr2xyxyxyxy(gt_bboxes)
# (b, n_boxes, 1, 2)
a, b, _, d = corners.split(1, dim=-2)
ab = b - a
ad = d - a
# (b, n_boxes, h*w, 2)
ap = xy_centers - a
norm_ab = (ab * ab).sum(dim=-1)
norm_ad = (ad * ad).sum(dim=-1)
ap_dot_ab = (ap * ab).sum(dim=-1)
ap_dot_ad = (ap * ad).sum(dim=-1)
return (ap_dot_ab >= 0) & (ap_dot_ab <= norm_ab) & (ap_dot_ad >= 0) & (ap_dot_ad <= norm_ad) # is_in_box
def make_anchors(feats, strides, grid_cell_offset=0.5):
"""Generate anchors from features."""
@ -255,7 +300,7 @@ def make_anchors(feats, strides, grid_cell_offset=0.5):
_, _, h, w = feats[i].shape
sx = torch.arange(end=w, device=device, dtype=dtype) + grid_cell_offset # shift x
sy = torch.arange(end=h, device=device, dtype=dtype) + grid_cell_offset # shift y
sy, sx = torch.meshgrid(sy, sx, indexing='ij') if TORCH_1_10 else torch.meshgrid(sy, sx)
sy, sx = torch.meshgrid(sy, sx, indexing="ij") if TORCH_1_10 else torch.meshgrid(sy, sx)
anchor_points.append(torch.stack((sx, sy), -1).view(-1, 2))
stride_tensor.append(torch.full((h * w, 1), stride, dtype=dtype, device=device))
return torch.cat(anchor_points), torch.cat(stride_tensor)
@ -263,7 +308,8 @@ def make_anchors(feats, strides, grid_cell_offset=0.5):
def dist2bbox(distance, anchor_points, xywh=True, dim=-1):
"""Transform distance(ltrb) to box(xywh or xyxy)."""
lt, rb = distance.chunk(2, dim)
assert(distance.shape[dim] == 4)
lt, rb = distance.split([2, 2], dim)
x1y1 = anchor_points - lt
x2y2 = anchor_points + rb
if xywh:
@ -277,3 +323,23 @@ def bbox2dist(anchor_points, bbox, reg_max):
"""Transform bbox(xyxy) to dist(ltrb)."""
x1y1, x2y2 = bbox.chunk(2, -1)
return torch.cat((anchor_points - x1y1, x2y2 - anchor_points), -1).clamp_(0, reg_max - 0.01) # dist (lt, rb)
def dist2rbox(pred_dist, pred_angle, anchor_points, dim=-1):
"""
Decode predicted object bounding box coordinates from anchor points and distribution.
Args:
pred_dist (torch.Tensor): Predicted rotated distance, (bs, h*w, 4).
pred_angle (torch.Tensor): Predicted angle, (bs, h*w, 1).
anchor_points (torch.Tensor): Anchor points, (h*w, 2).
Returns:
(torch.Tensor): Predicted rotated bounding boxes, (bs, h*w, 4).
"""
lt, rb = pred_dist.split(2, dim=dim)
cos, sin = torch.cos(pred_angle), torch.sin(pred_angle)
# (bs, h*w, 1)
xf, yf = ((rb - lt) / 2).split(1, dim=dim)
x, y = xf * cos - yf * sin, xf * sin + yf * cos
xy = torch.cat([x, y], dim=dim) + anchor_points
return torch.cat([xy, lt + rb], dim=dim)