ultralytics 8.2.38 official YOLOv10 support (#13113)
Signed-off-by: Glenn Jocher <glenn.jocher@ultralytics.com> Co-authored-by: UltralyticsAssistant <web@ultralytics.com> Co-authored-by: Laughing-q <1185102784@qq.com> Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com> Co-authored-by: Laughing <61612323+Laughing-q@users.noreply.github.com>
This commit is contained in:
Burhan
GitHub
parent
821e5fa477
commit
ffb46fd7fb
23 changed files with 785 additions and 32 deletions
|
|
@ -1,6 +1,7 @@
|
|||
# Ultralytics YOLO 🚀, AGPL-3.0 license
|
||||
"""Model head modules."""
|
||||
|
||||
import copy
|
||||
import math
|
||||
|
||||
import torch
|
||||
|
|
@ -14,7 +15,7 @@ from .conv import Conv
|
|||
from .transformer import MLP, DeformableTransformerDecoder, DeformableTransformerDecoderLayer
|
||||
from .utils import bias_init_with_prob, linear_init
|
||||
|
||||
__all__ = "Detect", "Segment", "Pose", "Classify", "OBB", "RTDETRDecoder"
|
||||
__all__ = "Detect", "Segment", "Pose", "Classify", "OBB", "RTDETRDecoder", "v10Detect"
|
||||
|
||||
|
||||
class Detect(nn.Module):
|
||||
|
|
@ -22,6 +23,8 @@ class Detect(nn.Module):
|
|||
|
||||
dynamic = False # force grid reconstruction
|
||||
export = False # export mode
|
||||
end2end = False # end2end
|
||||
max_det = 300 # max_det
|
||||
shape = None
|
||||
anchors = torch.empty(0) # init
|
||||
strides = torch.empty(0) # init
|
||||
|
|
@ -41,13 +44,48 @@ class Detect(nn.Module):
|
|||
self.cv3 = nn.ModuleList(nn.Sequential(Conv(x, c3, 3), Conv(c3, c3, 3), nn.Conv2d(c3, self.nc, 1)) for x in ch)
|
||||
self.dfl = DFL(self.reg_max) if self.reg_max > 1 else nn.Identity()
|
||||
|
||||
if self.end2end:
|
||||
self.one2one_cv2 = copy.deepcopy(self.cv2)
|
||||
self.one2one_cv3 = copy.deepcopy(self.cv3)
|
||||
|
||||
def forward(self, x):
|
||||
"""Concatenates and returns predicted bounding boxes and class probabilities."""
|
||||
if self.end2end:
|
||||
return self.forward_end2end(x)
|
||||
|
||||
for i in range(self.nl):
|
||||
x[i] = torch.cat((self.cv2[i](x[i]), self.cv3[i](x[i])), 1)
|
||||
if self.training: # Training path
|
||||
return x
|
||||
y = self._inference(x)
|
||||
return y if self.export else (y, x)
|
||||
|
||||
def forward_end2end(self, x):
|
||||
"""
|
||||
Performs forward pass of the v10Detect module.
|
||||
|
||||
Args:
|
||||
x (tensor): Input tensor.
|
||||
|
||||
Returns:
|
||||
(dict, tensor): If not in training mode, returns a dictionary containing the outputs of both one2many and one2one detections.
|
||||
If in training mode, returns a dictionary containing the outputs of one2many and one2one detections separately.
|
||||
"""
|
||||
x_detach = [xi.detach() for xi in x]
|
||||
one2one = [
|
||||
torch.cat((self.one2one_cv2[i](x_detach[i]), self.one2one_cv3[i](x_detach[i])), 1) for i in range(self.nl)
|
||||
]
|
||||
for i in range(self.nl):
|
||||
x[i] = torch.cat((self.cv2[i](x[i]), self.cv3[i](x[i])), 1)
|
||||
if self.training: # Training path
|
||||
return {"one2many": x, "one2one": one2one}
|
||||
|
||||
y = self._inference(one2one)
|
||||
y = self.postprocess(y.permute(0, 2, 1), self.max_det, self.nc)
|
||||
return y if self.export else (y, {"one2many": x, "one2one": one2one})
|
||||
|
||||
def _inference(self, x):
|
||||
"""Decode predicted bounding boxes and class probabilities based on multiple-level feature maps."""
|
||||
# Inference path
|
||||
shape = x[0].shape # BCHW
|
||||
x_cat = torch.cat([xi.view(shape[0], self.no, -1) for xi in x], 2)
|
||||
|
|
@ -73,7 +111,7 @@ class Detect(nn.Module):
|
|||
dbox = self.decode_bboxes(self.dfl(box), self.anchors.unsqueeze(0)) * self.strides
|
||||
|
||||
y = torch.cat((dbox, cls.sigmoid()), 1)
|
||||
return y if self.export else (y, x)
|
||||
return y
|
||||
|
||||
def bias_init(self):
|
||||
"""Initialize Detect() biases, WARNING: requires stride availability."""
|
||||
|
|
@ -83,10 +121,47 @@ class Detect(nn.Module):
|
|||
for a, b, s in zip(m.cv2, m.cv3, m.stride): # from
|
||||
a[-1].bias.data[:] = 1.0 # box
|
||||
b[-1].bias.data[: m.nc] = math.log(5 / m.nc / (640 / s) ** 2) # cls (.01 objects, 80 classes, 640 img)
|
||||
if self.end2end:
|
||||
for a, b, s in zip(m.one2one_cv2, m.one2one_cv3, m.stride): # from
|
||||
a[-1].bias.data[:] = 1.0 # box
|
||||
b[-1].bias.data[: m.nc] = math.log(5 / m.nc / (640 / s) ** 2) # cls (.01 objects, 80 classes, 640 img)
|
||||
|
||||
def decode_bboxes(self, bboxes, anchors):
|
||||
"""Decode bounding boxes."""
|
||||
return dist2bbox(bboxes, anchors, xywh=True, dim=1)
|
||||
return dist2bbox(bboxes, anchors, xywh=not self.end2end, dim=1)
|
||||
|
||||
@staticmethod
|
||||
def postprocess(preds: torch.Tensor, max_det: int, nc: int = 80):
|
||||
"""
|
||||
Post-processes the predictions obtained from a YOLOv10 model.
|
||||
|
||||
Args:
|
||||
preds (torch.Tensor): The predictions obtained from the model. It should have a shape of (batch_size, num_boxes, 4 + num_classes).
|
||||
max_det (int): The maximum number of detections to keep.
|
||||
nc (int, optional): The number of classes. Defaults to 80.
|
||||
|
||||
Returns:
|
||||
(torch.Tensor): The post-processed predictions with shape (batch_size, max_det, 6),
|
||||
including bounding boxes, scores and cls.
|
||||
"""
|
||||
assert 4 + nc == preds.shape[-1]
|
||||
boxes, scores = preds.split([4, nc], dim=-1)
|
||||
max_scores = scores.amax(dim=-1)
|
||||
max_scores, index = torch.topk(max_scores, min(max_det, max_scores.shape[1]), axis=-1)
|
||||
index = index.unsqueeze(-1)
|
||||
boxes = torch.gather(boxes, dim=1, index=index.repeat(1, 1, boxes.shape[-1]))
|
||||
scores = torch.gather(scores, dim=1, index=index.repeat(1, 1, scores.shape[-1]))
|
||||
|
||||
# NOTE: simplify but result slightly lower mAP
|
||||
# scores, labels = scores.max(dim=-1)
|
||||
# return torch.cat([boxes, scores.unsqueeze(-1), labels.unsqueeze(-1)], dim=-1)
|
||||
|
||||
scores, index = torch.topk(scores.flatten(1), max_det, axis=-1)
|
||||
labels = index % nc
|
||||
index = index // nc
|
||||
boxes = boxes.gather(dim=1, index=index.unsqueeze(-1).repeat(1, 1, boxes.shape[-1]))
|
||||
|
||||
return torch.cat([boxes, scores.unsqueeze(-1), labels.unsqueeze(-1).to(boxes.dtype)], dim=-1)
|
||||
|
||||
|
||||
class Segment(Detect):
|
||||
|
|
@ -487,3 +562,39 @@ class RTDETRDecoder(nn.Module):
|
|||
xavier_uniform_(self.query_pos_head.layers[1].weight)
|
||||
for layer in self.input_proj:
|
||||
xavier_uniform_(layer[0].weight)
|
||||
|
||||
|
||||
class v10Detect(Detect):
|
||||
"""
|
||||
v10 Detection head from https://arxiv.org/pdf/2405.14458
|
||||
|
||||
Args:
|
||||
nc (int): Number of classes.
|
||||
ch (tuple): Tuple of channel sizes.
|
||||
|
||||
Attributes:
|
||||
max_det (int): Maximum number of detections.
|
||||
|
||||
Methods:
|
||||
__init__(self, nc=80, ch=()): Initializes the v10Detect object.
|
||||
forward(self, x): Performs forward pass of the v10Detect module.
|
||||
bias_init(self): Initializes biases of the Detect module.
|
||||
|
||||
"""
|
||||
|
||||
end2end = True
|
||||
|
||||
def __init__(self, nc=80, ch=()):
|
||||
"""Initializes the v10Detect object with the specified number of classes and input channels."""
|
||||
super().__init__(nc, ch)
|
||||
c3 = max(ch[0], min(self.nc, 100)) # channels
|
||||
# Light cls head
|
||||
self.cv3 = nn.ModuleList(
|
||||
nn.Sequential(
|
||||
nn.Sequential(Conv(x, x, 3, g=x), Conv(x, c3, 1)),
|
||||
nn.Sequential(Conv(c3, c3, 3, g=c3), Conv(c3, c3, 1)),
|
||||
nn.Conv2d(c3, self.nc, 1),
|
||||
)
|
||||
for x in ch
|
||||
)
|
||||
self.one2one_cv3 = copy.deepcopy(self.cv3)
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue