ultralytics 8.0.98 add Baidu RT-DETR models (#2527)
Co-authored-by: Kalen Michael <kalenmike@gmail.com> Co-authored-by: Laughing <61612323+Laughing-q@users.noreply.github.com> Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> Co-authored-by: Yonghye Kwon <developer.0hye@gmail.com> Co-authored-by: Dowon <ks2515@naver.com>
This commit is contained in:
Glenn Jocher
GitHub
17
ultralytics/nn/modules/__init__.py
Normal file
17
ultralytics/nn/modules/__init__.py
Normal file
@ -0,0 +1,17 @@
|
||||
# Ultralytics YOLO 🚀, AGPL-3.0 license
|
||||
|
||||
from .block import (C1, C2, C3, C3TR, DFL, SPP, SPPF, Bottleneck, BottleneckCSP, C2f, C3Ghost, C3x, GhostBottleneck,
|
||||
HGBlock, HGStem, Proto, RepC3)
|
||||
from .conv import (CBAM, ChannelAttention, Concat, Conv, ConvTranspose, DWConv, DWConvTranspose2d, Focus, GhostConv,
|
||||
LightConv, RepConv, SpatialAttention)
|
||||
from .head import Classify, Detect, Pose, RTDETRDecoder, Segment
|
||||
from .transformer import (AIFI, MLP, DeformableTransformerDecoder, DeformableTransformerDecoderLayer, LayerNorm2d,
|
||||
MLPBlock, MSDeformAttn, TransformerBlock, TransformerEncoderLayer, TransformerLayer)
|
||||
|
||||
__all__ = [
|
||||
'Conv', 'LightConv', 'RepConv', 'DWConv', 'DWConvTranspose2d', 'ConvTranspose', 'Focus', 'GhostConv',
|
||||
'ChannelAttention', 'SpatialAttention', 'CBAM', 'Concat', 'TransformerLayer', 'TransformerBlock', 'MLPBlock',
|
||||
'LayerNorm2d', 'DFL', 'HGBlock', 'HGStem', 'SPP', 'SPPF', 'C1', 'C2', 'C3', 'C2f', 'C3x', 'C3TR', 'C3Ghost',
|
||||
'GhostBottleneck', 'Bottleneck', 'BottleneckCSP', 'Proto', 'Detect', 'Segment', 'Pose', 'Classify',
|
||||
'TransformerEncoderLayer', 'RepC3', 'RTDETRDecoder', 'AIFI', 'DeformableTransformerDecoder',
|
||||
'DeformableTransformerDecoderLayer', 'MSDeformAttn', 'MLP']
|
||||
305
ultralytics/nn/modules/block.py
Normal file
305
ultralytics/nn/modules/block.py
Normal file
@ -0,0 +1,305 @@
|
||||
# Ultralytics YOLO 🚀, AGPL-3.0 license
|
||||
"""
|
||||
Block modules
|
||||
"""
|
||||
|
||||
import torch
|
||||
import torch.nn as nn
|
||||
import torch.nn.functional as F
|
||||
|
||||
from .conv import Conv, DWConv, GhostConv, LightConv, RepConv
|
||||
from .transformer import TransformerBlock
|
||||
|
||||
__all__ = [
|
||||
'DFL', 'HGBlock', 'HGStem', 'SPP', 'SPPF', 'C1', 'C2', 'C3', 'C2f', 'C3x', 'C3TR', 'C3Ghost', 'GhostBottleneck',
|
||||
'Bottleneck', 'BottleneckCSP', 'Proto', 'RepC3']
|
||||
|
||||
|
||||
class DFL(nn.Module):
|
||||
"""
|
||||
Integral module of Distribution Focal Loss (DFL).
|
||||
Proposed in Generalized Focal Loss https://ieeexplore.ieee.org/document/9792391
|
||||
"""
|
||||
|
||||
def __init__(self, c1=16):
|
||||
"""Initialize a convolutional layer with a given number of input channels."""
|
||||
super().__init__()
|
||||
self.conv = nn.Conv2d(c1, 1, 1, bias=False).requires_grad_(False)
|
||||
x = torch.arange(c1, dtype=torch.float)
|
||||
self.conv.weight.data[:] = nn.Parameter(x.view(1, c1, 1, 1))
|
||||
self.c1 = c1
|
||||
|
||||
def forward(self, x):
|
||||
"""Applies a transformer layer on input tensor 'x' and returns a tensor."""
|
||||
b, c, a = x.shape # batch, channels, anchors
|
||||
return self.conv(x.view(b, 4, self.c1, a).transpose(2, 1).softmax(1)).view(b, 4, a)
|
||||
# return self.conv(x.view(b, self.c1, 4, a).softmax(1)).view(b, 4, a)
|
||||
|
||||
|
||||
class Proto(nn.Module):
|
||||
"""YOLOv8 mask Proto module for segmentation models."""
|
||||
|
||||
def __init__(self, c1, c_=256, c2=32): # ch_in, number of protos, number of masks
|
||||
super().__init__()
|
||||
self.cv1 = Conv(c1, c_, k=3)
|
||||
self.upsample = nn.ConvTranspose2d(c_, c_, 2, 2, 0, bias=True) # nn.Upsample(scale_factor=2, mode='nearest')
|
||||
self.cv2 = Conv(c_, c_, k=3)
|
||||
self.cv3 = Conv(c_, c2)
|
||||
|
||||
def forward(self, x):
|
||||
"""Performs a forward pass through layers using an upsampled input image."""
|
||||
return self.cv3(self.cv2(self.upsample(self.cv1(x))))
|
||||
|
||||
|
||||
class HGStem(nn.Module):
|
||||
"""StemBlock of PPHGNetV2 with 5 convolutions and one maxpool2d.
|
||||
https://github.com/PaddlePaddle/PaddleDetection/blob/develop/ppdet/modeling/backbones/hgnet_v2.py
|
||||
"""
|
||||
|
||||
def __init__(self, c1, cm, c2):
|
||||
super().__init__()
|
||||
self.stem1 = Conv(c1, cm, 3, 2, act=nn.ReLU())
|
||||
self.stem2a = Conv(cm, cm // 2, 2, 1, 0, act=nn.ReLU())
|
||||
self.stem2b = Conv(cm // 2, cm, 2, 1, 0, act=nn.ReLU())
|
||||
self.stem3 = Conv(cm * 2, cm, 3, 2, act=nn.ReLU())
|
||||
self.stem4 = Conv(cm, c2, 1, 1, act=nn.ReLU())
|
||||
self.pool = nn.MaxPool2d(kernel_size=2, stride=1, padding=0, ceil_mode=True)
|
||||
|
||||
def forward(self, x):
|
||||
"""Forward pass of a PPHGNetV2 backbone layer."""
|
||||
x = self.stem1(x)
|
||||
x = F.pad(x, [0, 1, 0, 1])
|
||||
x2 = self.stem2a(x)
|
||||
x2 = F.pad(x2, [0, 1, 0, 1])
|
||||
x2 = self.stem2b(x2)
|
||||
x1 = self.pool(x)
|
||||
x = torch.cat([x1, x2], dim=1)
|
||||
x = self.stem3(x)
|
||||
x = self.stem4(x)
|
||||
return x
|
||||
|
||||
|
||||
class HGBlock(nn.Module):
|
||||
"""HG_Block of PPHGNetV2 with 2 convolutions and LightConv.
|
||||
https://github.com/PaddlePaddle/PaddleDetection/blob/develop/ppdet/modeling/backbones/hgnet_v2.py
|
||||
"""
|
||||
|
||||
def __init__(self, c1, cm, c2, k=3, n=6, lightconv=False, shortcut=False, act=nn.ReLU()):
|
||||
super().__init__()
|
||||
block = LightConv if lightconv else Conv
|
||||
self.m = nn.ModuleList(block(c1 if i == 0 else cm, cm, k=k, act=act) for i in range(n))
|
||||
self.sc = Conv(c1 + n * cm, c2 // 2, 1, 1, act=act) # squeeze conv
|
||||
self.ec = Conv(c2 // 2, c2, 1, 1, act=act) # excitation conv
|
||||
self.add = shortcut and c1 == c2
|
||||
|
||||
def forward(self, x):
|
||||
"""Forward pass of a PPHGNetV2 backbone layer."""
|
||||
y = [x]
|
||||
y.extend(m(y[-1]) for m in self.m)
|
||||
y = self.ec(self.sc(torch.cat(y, 1)))
|
||||
return y + x if self.add else y
|
||||
|
||||
|
||||
class SPP(nn.Module):
|
||||
"""Spatial Pyramid Pooling (SPP) layer https://arxiv.org/abs/1406.4729."""
|
||||
|
||||
def __init__(self, c1, c2, k=(5, 9, 13)):
|
||||
"""Initialize the SPP layer with input/output channels and pooling kernel sizes."""
|
||||
super().__init__()
|
||||
c_ = c1 // 2 # hidden channels
|
||||
self.cv1 = Conv(c1, c_, 1, 1)
|
||||
self.cv2 = Conv(c_ * (len(k) + 1), c2, 1, 1)
|
||||
self.m = nn.ModuleList([nn.MaxPool2d(kernel_size=x, stride=1, padding=x // 2) for x in k])
|
||||
|
||||
def forward(self, x):
|
||||
"""Forward pass of the SPP layer, performing spatial pyramid pooling."""
|
||||
x = self.cv1(x)
|
||||
return self.cv2(torch.cat([x] + [m(x) for m in self.m], 1))
|
||||
|
||||
|
||||
class SPPF(nn.Module):
|
||||
"""Spatial Pyramid Pooling - Fast (SPPF) layer for YOLOv5 by Glenn Jocher."""
|
||||
|
||||
def __init__(self, c1, c2, k=5): # equivalent to SPP(k=(5, 9, 13))
|
||||
super().__init__()
|
||||
c_ = c1 // 2 # hidden channels
|
||||
self.cv1 = Conv(c1, c_, 1, 1)
|
||||
self.cv2 = Conv(c_ * 4, c2, 1, 1)
|
||||
self.m = nn.MaxPool2d(kernel_size=k, stride=1, padding=k // 2)
|
||||
|
||||
def forward(self, x):
|
||||
"""Forward pass through Ghost Convolution block."""
|
||||
x = self.cv1(x)
|
||||
y1 = self.m(x)
|
||||
y2 = self.m(y1)
|
||||
return self.cv2(torch.cat((x, y1, y2, self.m(y2)), 1))
|
||||
|
||||
|
||||
class C1(nn.Module):
|
||||
"""CSP Bottleneck with 1 convolution."""
|
||||
|
||||
def __init__(self, c1, c2, n=1): # ch_in, ch_out, number
|
||||
super().__init__()
|
||||
self.cv1 = Conv(c1, c2, 1, 1)
|
||||
self.m = nn.Sequential(*(Conv(c2, c2, 3) for _ in range(n)))
|
||||
|
||||
def forward(self, x):
|
||||
"""Applies cross-convolutions to input in the C3 module."""
|
||||
y = self.cv1(x)
|
||||
return self.m(y) + y
|
||||
|
||||
|
||||
class C2(nn.Module):
|
||||
"""CSP Bottleneck with 2 convolutions."""
|
||||
|
||||
def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5): # ch_in, ch_out, number, shortcut, groups, expansion
|
||||
super().__init__()
|
||||
self.c = int(c2 * e) # hidden channels
|
||||
self.cv1 = Conv(c1, 2 * self.c, 1, 1)
|
||||
self.cv2 = Conv(2 * self.c, c2, 1) # optional act=FReLU(c2)
|
||||
# self.attention = ChannelAttention(2 * self.c) # or SpatialAttention()
|
||||
self.m = nn.Sequential(*(Bottleneck(self.c, self.c, shortcut, g, k=((3, 3), (3, 3)), e=1.0) for _ in range(n)))
|
||||
|
||||
def forward(self, x):
|
||||
"""Forward pass through the CSP bottleneck with 2 convolutions."""
|
||||
a, b = self.cv1(x).chunk(2, 1)
|
||||
return self.cv2(torch.cat((self.m(a), b), 1))
|
||||
|
||||
|
||||
class C2f(nn.Module):
|
||||
"""CSP Bottleneck with 2 convolutions."""
|
||||
|
||||
def __init__(self, c1, c2, n=1, shortcut=False, g=1, e=0.5): # ch_in, ch_out, number, shortcut, groups, expansion
|
||||
super().__init__()
|
||||
self.c = int(c2 * e) # hidden channels
|
||||
self.cv1 = Conv(c1, 2 * self.c, 1, 1)
|
||||
self.cv2 = Conv((2 + n) * self.c, c2, 1) # optional act=FReLU(c2)
|
||||
self.m = nn.ModuleList(Bottleneck(self.c, self.c, shortcut, g, k=((3, 3), (3, 3)), e=1.0) for _ in range(n))
|
||||
|
||||
def forward(self, x):
|
||||
"""Forward pass of a YOLOv5 CSPDarknet backbone layer."""
|
||||
y = list(self.cv1(x).chunk(2, 1))
|
||||
y.extend(m(y[-1]) for m in self.m)
|
||||
return self.cv2(torch.cat(y, 1))
|
||||
|
||||
def forward_split(self, x):
|
||||
"""Applies spatial attention to module's input."""
|
||||
y = list(self.cv1(x).split((self.c, self.c), 1))
|
||||
y.extend(m(y[-1]) for m in self.m)
|
||||
return self.cv2(torch.cat(y, 1))
|
||||
|
||||
|
||||
class C3(nn.Module):
|
||||
"""CSP Bottleneck with 3 convolutions."""
|
||||
|
||||
def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5): # ch_in, ch_out, number, shortcut, groups, expansion
|
||||
super().__init__()
|
||||
c_ = int(c2 * e) # hidden channels
|
||||
self.cv1 = Conv(c1, c_, 1, 1)
|
||||
self.cv2 = Conv(c1, c_, 1, 1)
|
||||
self.cv3 = Conv(2 * c_, c2, 1) # optional act=FReLU(c2)
|
||||
self.m = nn.Sequential(*(Bottleneck(c_, c_, shortcut, g, k=((1, 1), (3, 3)), e=1.0) for _ in range(n)))
|
||||
|
||||
def forward(self, x):
|
||||
"""Forward pass through the CSP bottleneck with 2 convolutions."""
|
||||
return self.cv3(torch.cat((self.m(self.cv1(x)), self.cv2(x)), 1))
|
||||
|
||||
|
||||
class C3x(C3):
|
||||
"""C3 module with cross-convolutions."""
|
||||
|
||||
def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5):
|
||||
"""Initialize C3TR instance and set default parameters."""
|
||||
super().__init__(c1, c2, n, shortcut, g, e)
|
||||
self.c_ = int(c2 * e)
|
||||
self.m = nn.Sequential(*(Bottleneck(self.c_, self.c_, shortcut, g, k=((1, 3), (3, 1)), e=1) for _ in range(n)))
|
||||
|
||||
|
||||
class RepC3(nn.Module):
|
||||
"""Rep C3."""
|
||||
|
||||
def __init__(self, c1, c2, n=3, e=1.0):
|
||||
super().__init__()
|
||||
c_ = int(c2 * e) # hidden channels
|
||||
self.cv1 = Conv(c1, c2, 1, 1)
|
||||
self.cv2 = Conv(c1, c2, 1, 1)
|
||||
self.m = nn.Sequential(*[RepConv(c_, c_) for _ in range(n)])
|
||||
self.cv3 = Conv(c_, c2, 1, 1) if c_ != c2 else nn.Identity()
|
||||
|
||||
def forward(self, x):
|
||||
"""Forward pass of RT-DETR neck layer."""
|
||||
return self.cv3(self.m(self.cv1(x)) + self.cv2(x))
|
||||
|
||||
|
||||
class C3TR(C3):
|
||||
"""C3 module with TransformerBlock()."""
|
||||
|
||||
def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5):
|
||||
"""Initialize C3Ghost module with GhostBottleneck()."""
|
||||
super().__init__(c1, c2, n, shortcut, g, e)
|
||||
c_ = int(c2 * e)
|
||||
self.m = TransformerBlock(c_, c_, 4, n)
|
||||
|
||||
|
||||
class C3Ghost(C3):
|
||||
"""C3 module with GhostBottleneck()."""
|
||||
|
||||
def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5):
|
||||
"""Initialize 'SPP' module with various pooling sizes for spatial pyramid pooling."""
|
||||
super().__init__(c1, c2, n, shortcut, g, e)
|
||||
c_ = int(c2 * e) # hidden channels
|
||||
self.m = nn.Sequential(*(GhostBottleneck(c_, c_) for _ in range(n)))
|
||||
|
||||
|
||||
class GhostBottleneck(nn.Module):
|
||||
"""Ghost Bottleneck https://github.com/huawei-noah/ghostnet."""
|
||||
|
||||
def __init__(self, c1, c2, k=3, s=1): # ch_in, ch_out, kernel, stride
|
||||
super().__init__()
|
||||
c_ = c2 // 2
|
||||
self.conv = nn.Sequential(
|
||||
GhostConv(c1, c_, 1, 1), # pw
|
||||
DWConv(c_, c_, k, s, act=False) if s == 2 else nn.Identity(), # dw
|
||||
GhostConv(c_, c2, 1, 1, act=False)) # pw-linear
|
||||
self.shortcut = nn.Sequential(DWConv(c1, c1, k, s, act=False), Conv(c1, c2, 1, 1,
|
||||
act=False)) if s == 2 else nn.Identity()
|
||||
|
||||
def forward(self, x):
|
||||
"""Applies skip connection and concatenation to input tensor."""
|
||||
return self.conv(x) + self.shortcut(x)
|
||||
|
||||
|
||||
class Bottleneck(nn.Module):
|
||||
"""Standard bottleneck."""
|
||||
|
||||
def __init__(self, c1, c2, shortcut=True, g=1, k=(3, 3), e=0.5): # ch_in, ch_out, shortcut, groups, kernels, expand
|
||||
super().__init__()
|
||||
c_ = int(c2 * e) # hidden channels
|
||||
self.cv1 = Conv(c1, c_, k[0], 1)
|
||||
self.cv2 = Conv(c_, c2, k[1], 1, g=g)
|
||||
self.add = shortcut and c1 == c2
|
||||
|
||||
def forward(self, x):
|
||||
"""'forward()' applies the YOLOv5 FPN to input data."""
|
||||
return x + self.cv2(self.cv1(x)) if self.add else self.cv2(self.cv1(x))
|
||||
|
||||
|
||||
class BottleneckCSP(nn.Module):
|
||||
"""CSP Bottleneck https://github.com/WongKinYiu/CrossStagePartialNetworks."""
|
||||
|
||||
def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5): # ch_in, ch_out, number, shortcut, groups, expansion
|
||||
super().__init__()
|
||||
c_ = int(c2 * e) # hidden channels
|
||||
self.cv1 = Conv(c1, c_, 1, 1)
|
||||
self.cv2 = nn.Conv2d(c1, c_, 1, 1, bias=False)
|
||||
self.cv3 = nn.Conv2d(c_, c_, 1, 1, bias=False)
|
||||
self.cv4 = Conv(2 * c_, c2, 1, 1)
|
||||
self.bn = nn.BatchNorm2d(2 * c_) # applied to cat(cv2, cv3)
|
||||
self.act = nn.SiLU()
|
||||
self.m = nn.Sequential(*(Bottleneck(c_, c_, shortcut, g, e=1.0) for _ in range(n)))
|
||||
|
||||
def forward(self, x):
|
||||
"""Applies a CSP bottleneck with 3 convolutions."""
|
||||
y1 = self.cv3(self.m(self.cv1(x)))
|
||||
y2 = self.cv2(x)
|
||||
return self.cv4(self.act(self.bn(torch.cat((y1, y2), 1))))
|
||||
277
ultralytics/nn/modules/conv.py
Normal file
277
ultralytics/nn/modules/conv.py
Normal file
@ -0,0 +1,277 @@
|
||||
# Ultralytics YOLO 🚀, AGPL-3.0 license
|
||||
"""
|
||||
Convolution modules
|
||||
"""
|
||||
|
||||
import math
|
||||
|
||||
import numpy as np
|
||||
import torch
|
||||
import torch.nn as nn
|
||||
|
||||
__all__ = [
|
||||
'Conv', 'LightConv', 'DWConv', 'DWConvTranspose2d', 'ConvTranspose', 'Focus', 'GhostConv', 'ChannelAttention',
|
||||
'SpatialAttention', 'CBAM', 'Concat', 'RepConv']
|
||||
|
||||
|
||||
def autopad(k, p=None, d=1): # kernel, padding, dilation
|
||||
"""Pad to 'same' shape outputs."""
|
||||
if d > 1:
|
||||
k = d * (k - 1) + 1 if isinstance(k, int) else [d * (x - 1) + 1 for x in k] # actual kernel-size
|
||||
if p is None:
|
||||
p = k // 2 if isinstance(k, int) else [x // 2 for x in k] # auto-pad
|
||||
return p
|
||||
|
||||
|
||||
class Conv(nn.Module):
|
||||
"""Standard convolution with args(ch_in, ch_out, kernel, stride, padding, groups, dilation, activation)."""
|
||||
default_act = nn.SiLU() # default activation
|
||||
|
||||
def __init__(self, c1, c2, k=1, s=1, p=None, g=1, d=1, act=True):
|
||||
"""Initialize Conv layer with given arguments including activation."""
|
||||
super().__init__()
|
||||
self.conv = nn.Conv2d(c1, c2, k, s, autopad(k, p, d), groups=g, dilation=d, bias=False)
|
||||
self.bn = nn.BatchNorm2d(c2)
|
||||
self.act = self.default_act if act is True else act if isinstance(act, nn.Module) else nn.Identity()
|
||||
|
||||
def forward(self, x):
|
||||
"""Apply convolution, batch normalization and activation to input tensor."""
|
||||
return self.act(self.bn(self.conv(x)))
|
||||
|
||||
def forward_fuse(self, x):
|
||||
"""Perform transposed convolution of 2D data."""
|
||||
return self.act(self.conv(x))
|
||||
|
||||
|
||||
class LightConv(nn.Module):
|
||||
"""Light convolution with args(ch_in, ch_out, kernel).
|
||||
https://github.com/PaddlePaddle/PaddleDetection/blob/develop/ppdet/modeling/backbones/hgnet_v2.py
|
||||
"""
|
||||
|
||||
def __init__(self, c1, c2, k=1, act=nn.ReLU()):
|
||||
"""Initialize Conv layer with given arguments including activation."""
|
||||
super().__init__()
|
||||
self.conv1 = Conv(c1, c2, 1, act=False)
|
||||
self.conv2 = DWConv(c2, c2, k, act=act)
|
||||
|
||||
def forward(self, x):
|
||||
"""Apply 2 convolutions to input tensor."""
|
||||
return self.conv2(self.conv1(x))
|
||||
|
||||
|
||||
class DWConv(Conv):
|
||||
"""Depth-wise convolution."""
|
||||
|
||||
def __init__(self, c1, c2, k=1, s=1, d=1, act=True): # ch_in, ch_out, kernel, stride, dilation, activation
|
||||
super().__init__(c1, c2, k, s, g=math.gcd(c1, c2), d=d, act=act)
|
||||
|
||||
|
||||
class DWConvTranspose2d(nn.ConvTranspose2d):
|
||||
"""Depth-wise transpose convolution."""
|
||||
|
||||
def __init__(self, c1, c2, k=1, s=1, p1=0, p2=0): # ch_in, ch_out, kernel, stride, padding, padding_out
|
||||
super().__init__(c1, c2, k, s, p1, p2, groups=math.gcd(c1, c2))
|
||||
|
||||
|
||||
class ConvTranspose(nn.Module):
|
||||
"""Convolution transpose 2d layer."""
|
||||
default_act = nn.SiLU() # default activation
|
||||
|
||||
def __init__(self, c1, c2, k=2, s=2, p=0, bn=True, act=True):
|
||||
"""Initialize ConvTranspose2d layer with batch normalization and activation function."""
|
||||
super().__init__()
|
||||
self.conv_transpose = nn.ConvTranspose2d(c1, c2, k, s, p, bias=not bn)
|
||||
self.bn = nn.BatchNorm2d(c2) if bn else nn.Identity()
|
||||
self.act = self.default_act if act is True else act if isinstance(act, nn.Module) else nn.Identity()
|
||||
|
||||
def forward(self, x):
|
||||
"""Applies transposed convolutions, batch normalization and activation to input."""
|
||||
return self.act(self.bn(self.conv_transpose(x)))
|
||||
|
||||
def forward_fuse(self, x):
|
||||
"""Applies activation and convolution transpose operation to input."""
|
||||
return self.act(self.conv_transpose(x))
|
||||
|
||||
|
||||
class Focus(nn.Module):
|
||||
"""Focus wh information into c-space."""
|
||||
|
||||
def __init__(self, c1, c2, k=1, s=1, p=None, g=1, act=True): # ch_in, ch_out, kernel, stride, padding, groups
|
||||
super().__init__()
|
||||
self.conv = Conv(c1 * 4, c2, k, s, p, g, act=act)
|
||||
# self.contract = Contract(gain=2)
|
||||
|
||||
def forward(self, x): # x(b,c,w,h) -> y(b,4c,w/2,h/2)
|
||||
return self.conv(torch.cat((x[..., ::2, ::2], x[..., 1::2, ::2], x[..., ::2, 1::2], x[..., 1::2, 1::2]), 1))
|
||||
# return self.conv(self.contract(x))
|
||||
|
||||
|
||||
class GhostConv(nn.Module):
|
||||
"""Ghost Convolution https://github.com/huawei-noah/ghostnet."""
|
||||
|
||||
def __init__(self, c1, c2, k=1, s=1, g=1, act=True): # ch_in, ch_out, kernel, stride, groups
|
||||
super().__init__()
|
||||
c_ = c2 // 2 # hidden channels
|
||||
self.cv1 = Conv(c1, c_, k, s, None, g, act=act)
|
||||
self.cv2 = Conv(c_, c_, 5, 1, None, c_, act=act)
|
||||
|
||||
def forward(self, x):
|
||||
"""Forward propagation through a Ghost Bottleneck layer with skip connection."""
|
||||
y = self.cv1(x)
|
||||
return torch.cat((y, self.cv2(y)), 1)
|
||||
|
||||
|
||||
class RepConv(nn.Module):
|
||||
"""RepConv is a basic rep-style block, including training and deploy status
|
||||
This code is based on https://github.com/DingXiaoH/RepVGG/blob/main/repvgg.py
|
||||
"""
|
||||
default_act = nn.SiLU() # default activation
|
||||
|
||||
def __init__(self, c1, c2, k=3, s=1, p=1, g=1, d=1, act=True, bn=False, deploy=False):
|
||||
super().__init__()
|
||||
assert k == 3 and p == 1
|
||||
self.g = g
|
||||
self.c1 = c1
|
||||
self.c2 = c2
|
||||
self.act = self.default_act if act is True else act if isinstance(act, nn.Module) else nn.Identity()
|
||||
|
||||
self.bn = nn.BatchNorm2d(num_features=c1) if bn and c2 == c1 and s == 1 else None
|
||||
self.conv1 = Conv(c1, c2, k, s, p=p, g=g, act=False)
|
||||
self.conv2 = Conv(c1, c2, 1, s, p=(p - k // 2), g=g, act=False)
|
||||
|
||||
def forward_fuse(self, x):
|
||||
"""Forward process"""
|
||||
return self.act(self.conv(x))
|
||||
|
||||
def forward(self, x):
|
||||
"""Forward process"""
|
||||
id_out = 0 if self.bn is None else self.bn(x)
|
||||
return self.act(self.conv1(x) + self.conv2(x) + id_out)
|
||||
|
||||
def get_equivalent_kernel_bias(self):
|
||||
kernel3x3, bias3x3 = self._fuse_bn_tensor(self.conv1)
|
||||
kernel1x1, bias1x1 = self._fuse_bn_tensor(self.conv2)
|
||||
kernelid, biasid = self._fuse_bn_tensor(self.bn)
|
||||
return kernel3x3 + self._pad_1x1_to_3x3_tensor(kernel1x1) + kernelid, bias3x3 + bias1x1 + biasid
|
||||
|
||||
def _avg_to_3x3_tensor(self, avgp):
|
||||
channels = self.c1
|
||||
groups = self.g
|
||||
kernel_size = avgp.kernel_size
|
||||
input_dim = channels // groups
|
||||
k = torch.zeros((channels, input_dim, kernel_size, kernel_size))
|
||||
k[np.arange(channels), np.tile(np.arange(input_dim), groups), :, :] = 1.0 / kernel_size ** 2
|
||||
return k
|
||||
|
||||
def _pad_1x1_to_3x3_tensor(self, kernel1x1):
|
||||
if kernel1x1 is None:
|
||||
return 0
|
||||
else:
|
||||
return torch.nn.functional.pad(kernel1x1, [1, 1, 1, 1])
|
||||
|
||||
def _fuse_bn_tensor(self, branch):
|
||||
if branch is None:
|
||||
return 0, 0
|
||||
if isinstance(branch, Conv):
|
||||
kernel = branch.conv.weight
|
||||
running_mean = branch.bn.running_mean
|
||||
running_var = branch.bn.running_var
|
||||
gamma = branch.bn.weight
|
||||
beta = branch.bn.bias
|
||||
eps = branch.bn.eps
|
||||
elif isinstance(branch, nn.BatchNorm2d):
|
||||
if not hasattr(self, 'id_tensor'):
|
||||
input_dim = self.c1 // self.g
|
||||
kernel_value = np.zeros((self.c1, input_dim, 3, 3), dtype=np.float32)
|
||||
for i in range(self.c1):
|
||||
kernel_value[i, i % input_dim, 1, 1] = 1
|
||||
self.id_tensor = torch.from_numpy(kernel_value).to(branch.weight.device)
|
||||
kernel = self.id_tensor
|
||||
running_mean = branch.running_mean
|
||||
running_var = branch.running_var
|
||||
gamma = branch.weight
|
||||
beta = branch.bias
|
||||
eps = branch.eps
|
||||
std = (running_var + eps).sqrt()
|
||||
t = (gamma / std).reshape(-1, 1, 1, 1)
|
||||
return kernel * t, beta - running_mean * gamma / std
|
||||
|
||||
def fuse_convs(self):
|
||||
if hasattr(self, 'conv'):
|
||||
return
|
||||
kernel, bias = self.get_equivalent_kernel_bias()
|
||||
self.conv = nn.Conv2d(in_channels=self.conv1.conv.in_channels,
|
||||
out_channels=self.conv1.conv.out_channels,
|
||||
kernel_size=self.conv1.conv.kernel_size,
|
||||
stride=self.conv1.conv.stride,
|
||||
padding=self.conv1.conv.padding,
|
||||
dilation=self.conv1.conv.dilation,
|
||||
groups=self.conv1.conv.groups,
|
||||
bias=True).requires_grad_(False)
|
||||
self.conv.weight.data = kernel
|
||||
self.conv.bias.data = bias
|
||||
for para in self.parameters():
|
||||
para.detach_()
|
||||
self.__delattr__('conv1')
|
||||
self.__delattr__('conv2')
|
||||
if hasattr(self, 'nm'):
|
||||
self.__delattr__('nm')
|
||||
if hasattr(self, 'bn'):
|
||||
self.__delattr__('bn')
|
||||
if hasattr(self, 'id_tensor'):
|
||||
self.__delattr__('id_tensor')
|
||||
|
||||
|
||||
class ChannelAttention(nn.Module):
|
||||
"""Channel-attention module https://github.com/open-mmlab/mmdetection/tree/v3.0.0rc1/configs/rtmdet."""
|
||||
|
||||
def __init__(self, channels: int) -> None:
|
||||
super().__init__()
|
||||
self.pool = nn.AdaptiveAvgPool2d(1)
|
||||
self.fc = nn.Conv2d(channels, channels, 1, 1, 0, bias=True)
|
||||
self.act = nn.Sigmoid()
|
||||
|
||||
def forward(self, x: torch.Tensor) -> torch.Tensor:
|
||||
return x * self.act(self.fc(self.pool(x)))
|
||||
|
||||
|
||||
class SpatialAttention(nn.Module):
|
||||
"""Spatial-attention module."""
|
||||
|
||||
def __init__(self, kernel_size=7):
|
||||
"""Initialize Spatial-attention module with kernel size argument."""
|
||||
super().__init__()
|
||||
assert kernel_size in (3, 7), 'kernel size must be 3 or 7'
|
||||
padding = 3 if kernel_size == 7 else 1
|
||||
self.cv1 = nn.Conv2d(2, 1, kernel_size, padding=padding, bias=False)
|
||||
self.act = nn.Sigmoid()
|
||||
|
||||
def forward(self, x):
|
||||
"""Apply channel and spatial attention on input for feature recalibration."""
|
||||
return x * self.act(self.cv1(torch.cat([torch.mean(x, 1, keepdim=True), torch.max(x, 1, keepdim=True)[0]], 1)))
|
||||
|
||||
|
||||
class CBAM(nn.Module):
|
||||
"""Convolutional Block Attention Module."""
|
||||
|
||||
def __init__(self, c1, kernel_size=7): # ch_in, kernels
|
||||
super().__init__()
|
||||
self.channel_attention = ChannelAttention(c1)
|
||||
self.spatial_attention = SpatialAttention(kernel_size)
|
||||
|
||||
def forward(self, x):
|
||||
"""Applies the forward pass through C1 module."""
|
||||
return self.spatial_attention(self.channel_attention(x))
|
||||
|
||||
|
||||
class Concat(nn.Module):
|
||||
"""Concatenate a list of tensors along dimension."""
|
||||
|
||||
def __init__(self, dimension=1):
|
||||
"""Concatenates a list of tensors along a specified dimension."""
|
||||
super().__init__()
|
||||
self.d = dimension
|
||||
|
||||
def forward(self, x):
|
||||
"""Forward pass for the YOLOv8 mask Proto module."""
|
||||
return torch.cat(x, self.d)
|
||||
382
ultralytics/nn/modules/head.py
Normal file
382
ultralytics/nn/modules/head.py
Normal file
@ -0,0 +1,382 @@
|
||||
# Ultralytics YOLO 🚀, AGPL-3.0 license
|
||||
"""
|
||||
Model head modules
|
||||
"""
|
||||
|
||||
import math
|
||||
|
||||
import torch
|
||||
import torch.nn as nn
|
||||
from torch.nn.init import constant_, xavier_uniform_
|
||||
|
||||
from ultralytics.yolo.utils.tal import dist2bbox, make_anchors
|
||||
|
||||
from .block import DFL, Proto
|
||||
from .conv import Conv
|
||||
from .transformer import MLP, DeformableTransformerDecoder, DeformableTransformerDecoderLayer
|
||||
from .utils import bias_init_with_prob, linear_init_
|
||||
|
||||
__all__ = ['Detect', 'Segment', 'Pose', 'Classify', 'RTDETRDecoder']
|
||||
|
||||
|
||||
class Detect(nn.Module):
|
||||
"""YOLOv8 Detect head for detection models."""
|
||||
dynamic = False # force grid reconstruction
|
||||
export = False # export mode
|
||||
shape = None
|
||||
anchors = torch.empty(0) # init
|
||||
strides = torch.empty(0) # init
|
||||
|
||||
def __init__(self, nc=80, ch=()): # detection layer
|
||||
super().__init__()
|
||||
self.nc = nc # number of classes
|
||||
self.nl = len(ch) # number of detection layers
|
||||
self.reg_max = 16 # DFL channels (ch[0] // 16 to scale 4/8/12/16/20 for n/s/m/l/x)
|
||||
self.no = nc + self.reg_max * 4 # number of outputs per anchor
|
||||
self.stride = torch.zeros(self.nl) # strides computed during build
|
||||
c2, c3 = max((16, ch[0] // 4, self.reg_max * 4)), max(ch[0], self.nc) # channels
|
||||
self.cv2 = nn.ModuleList(
|
||||
nn.Sequential(Conv(x, c2, 3), Conv(c2, c2, 3), nn.Conv2d(c2, 4 * self.reg_max, 1)) for x in ch)
|
||||
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()
|
||||
|
||||
def forward(self, x):
|
||||
"""Concatenates and returns predicted bounding boxes and class probabilities."""
|
||||
shape = x[0].shape # BCHW
|
||||
for i in range(self.nl):
|
||||
x[i] = torch.cat((self.cv2[i](x[i]), self.cv3[i](x[i])), 1)
|
||||
if self.training:
|
||||
return x
|
||||
elif self.dynamic or self.shape != shape:
|
||||
self.anchors, self.strides = (x.transpose(0, 1) for x in make_anchors(x, self.stride, 0.5))
|
||||
self.shape = shape
|
||||
|
||||
x_cat = torch.cat([xi.view(shape[0], self.no, -1) for xi in x], 2)
|
||||
if self.export and self.format in ('saved_model', 'pb', 'tflite', 'edgetpu', 'tfjs'): # avoid TF FlexSplitV ops
|
||||
box = x_cat[:, :self.reg_max * 4]
|
||||
cls = x_cat[:, self.reg_max * 4:]
|
||||
else:
|
||||
box, cls = x_cat.split((self.reg_max * 4, self.nc), 1)
|
||||
dbox = dist2bbox(self.dfl(box), self.anchors.unsqueeze(0), xywh=True, dim=1) * self.strides
|
||||
y = torch.cat((dbox, cls.sigmoid()), 1)
|
||||
return y if self.export else (y, x)
|
||||
|
||||
def bias_init(self):
|
||||
"""Initialize Detect() biases, WARNING: requires stride availability."""
|
||||
m = self # self.model[-1] # Detect() module
|
||||
# cf = torch.bincount(torch.tensor(np.concatenate(dataset.labels, 0)[:, 0]).long(), minlength=nc) + 1
|
||||
# ncf = math.log(0.6 / (m.nc - 0.999999)) if cf is None else torch.log(cf / cf.sum()) # nominal class frequency
|
||||
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)
|
||||
|
||||
|
||||
class Segment(Detect):
|
||||
"""YOLOv8 Segment head for segmentation models."""
|
||||
|
||||
def __init__(self, nc=80, nm=32, npr=256, ch=()):
|
||||
"""Initialize the YOLO model attributes such as the number of masks, prototypes, and the convolution layers."""
|
||||
super().__init__(nc, ch)
|
||||
self.nm = nm # number of masks
|
||||
self.npr = npr # number of protos
|
||||
self.proto = Proto(ch[0], self.npr, self.nm) # protos
|
||||
self.detect = Detect.forward
|
||||
|
||||
c4 = max(ch[0] // 4, self.nm)
|
||||
self.cv4 = nn.ModuleList(nn.Sequential(Conv(x, c4, 3), Conv(c4, c4, 3), nn.Conv2d(c4, self.nm, 1)) for x in ch)
|
||||
|
||||
def forward(self, x):
|
||||
"""Return model outputs and mask coefficients if training, otherwise return outputs and mask coefficients."""
|
||||
p = self.proto(x[0]) # mask protos
|
||||
bs = p.shape[0] # batch size
|
||||
|
||||
mc = torch.cat([self.cv4[i](x[i]).view(bs, self.nm, -1) for i in range(self.nl)], 2) # mask coefficients
|
||||
x = self.detect(self, x)
|
||||
if self.training:
|
||||
return x, mc, p
|
||||
return (torch.cat([x, mc], 1), p) if self.export else (torch.cat([x[0], mc], 1), (x[1], mc, p))
|
||||
|
||||
|
||||
class Pose(Detect):
|
||||
"""YOLOv8 Pose head for keypoints models."""
|
||||
|
||||
def __init__(self, nc=80, kpt_shape=(17, 3), ch=()):
|
||||
"""Initialize YOLO network with default parameters and Convolutional Layers."""
|
||||
super().__init__(nc, ch)
|
||||
self.kpt_shape = kpt_shape # number of keypoints, number of dims (2 for x,y or 3 for x,y,visible)
|
||||
self.nk = kpt_shape[0] * kpt_shape[1] # number of keypoints total
|
||||
self.detect = Detect.forward
|
||||
|
||||
c4 = max(ch[0] // 4, self.nk)
|
||||
self.cv4 = nn.ModuleList(nn.Sequential(Conv(x, c4, 3), Conv(c4, c4, 3), nn.Conv2d(c4, self.nk, 1)) for x in ch)
|
||||
|
||||
def forward(self, x):
|
||||
"""Perform forward pass through YOLO model and return predictions."""
|
||||
bs = x[0].shape[0] # batch size
|
||||
kpt = torch.cat([self.cv4[i](x[i]).view(bs, self.nk, -1) for i in range(self.nl)], -1) # (bs, 17*3, h*w)
|
||||
x = self.detect(self, x)
|
||||
if self.training:
|
||||
return x, kpt
|
||||
pred_kpt = self.kpts_decode(bs, kpt)
|
||||
return torch.cat([x, pred_kpt], 1) if self.export else (torch.cat([x[0], pred_kpt], 1), (x[1], kpt))
|
||||
|
||||
def kpts_decode(self, bs, kpts):
|
||||
"""Decodes keypoints."""
|
||||
ndim = self.kpt_shape[1]
|
||||
if self.export: # required for TFLite export to avoid 'PLACEHOLDER_FOR_GREATER_OP_CODES' bug
|
||||
y = kpts.view(bs, *self.kpt_shape, -1)
|
||||
a = (y[:, :, :2] * 2.0 + (self.anchors - 0.5)) * self.strides
|
||||
if ndim == 3:
|
||||
a = torch.cat((a, y[:, :, 1:2].sigmoid()), 2)
|
||||
return a.view(bs, self.nk, -1)
|
||||
else:
|
||||
y = kpts.clone()
|
||||
if ndim == 3:
|
||||
y[:, 2::3].sigmoid_() # inplace sigmoid
|
||||
y[:, 0::ndim] = (y[:, 0::ndim] * 2.0 + (self.anchors[0] - 0.5)) * self.strides
|
||||
y[:, 1::ndim] = (y[:, 1::ndim] * 2.0 + (self.anchors[1] - 0.5)) * self.strides
|
||||
return y
|
||||
|
||||
|
||||
class Classify(nn.Module):
|
||||
"""YOLOv8 classification head, i.e. x(b,c1,20,20) to x(b,c2)."""
|
||||
|
||||
def __init__(self, c1, c2, k=1, s=1, p=None, g=1): # ch_in, ch_out, kernel, stride, padding, groups
|
||||
super().__init__()
|
||||
c_ = 1280 # efficientnet_b0 size
|
||||
self.conv = Conv(c1, c_, k, s, p, g)
|
||||
self.pool = nn.AdaptiveAvgPool2d(1) # to x(b,c_,1,1)
|
||||
self.drop = nn.Dropout(p=0.0, inplace=True)
|
||||
self.linear = nn.Linear(c_, c2) # to x(b,c2)
|
||||
|
||||
def forward(self, x):
|
||||
"""Performs a forward pass of the YOLO model on input image data."""
|
||||
if isinstance(x, list):
|
||||
x = torch.cat(x, 1)
|
||||
x = self.linear(self.drop(self.pool(self.conv(x)).flatten(1)))
|
||||
return x if self.training else x.softmax(1)
|
||||
|
||||
|
||||
class RTDETRDecoder(nn.Module):
|
||||
|
||||
def __init__(
|
||||
self,
|
||||
nc=80,
|
||||
ch=(512, 1024, 2048),
|
||||
hidden_dim=256,
|
||||
num_queries=300,
|
||||
strides=(8, 16, 32), # TODO
|
||||
nl=3,
|
||||
num_decoder_points=4,
|
||||
nhead=8,
|
||||
num_decoder_layers=6,
|
||||
dim_feedforward=1024,
|
||||
dropout=0.,
|
||||
act=nn.ReLU(),
|
||||
eval_idx=-1,
|
||||
# training args
|
||||
num_denoising=100,
|
||||
label_noise_ratio=0.5,
|
||||
box_noise_scale=1.0,
|
||||
learnt_init_query=False):
|
||||
super().__init__()
|
||||
assert len(ch) <= nl
|
||||
assert len(strides) == len(ch)
|
||||
for _ in range(nl - len(strides)):
|
||||
strides.append(strides[-1] * 2)
|
||||
|
||||
self.hidden_dim = hidden_dim
|
||||
self.nhead = nhead
|
||||
self.feat_strides = strides
|
||||
self.nl = nl
|
||||
self.nc = nc
|
||||
self.num_queries = num_queries
|
||||
self.num_decoder_layers = num_decoder_layers
|
||||
|
||||
# backbone feature projection
|
||||
self._build_input_proj_layer(ch)
|
||||
|
||||
# Transformer module
|
||||
decoder_layer = DeformableTransformerDecoderLayer(hidden_dim, nhead, dim_feedforward, dropout, act, nl,
|
||||
num_decoder_points)
|
||||
self.decoder = DeformableTransformerDecoder(hidden_dim, decoder_layer, num_decoder_layers, eval_idx)
|
||||
|
||||
# denoising part
|
||||
self.denoising_class_embed = nn.Embedding(nc, hidden_dim)
|
||||
self.num_denoising = num_denoising
|
||||
self.label_noise_ratio = label_noise_ratio
|
||||
self.box_noise_scale = box_noise_scale
|
||||
|
||||
# decoder embedding
|
||||
self.learnt_init_query = learnt_init_query
|
||||
if learnt_init_query:
|
||||
self.tgt_embed = nn.Embedding(num_queries, hidden_dim)
|
||||
self.query_pos_head = MLP(4, 2 * hidden_dim, hidden_dim, num_layers=2)
|
||||
|
||||
# encoder head
|
||||
self.enc_output = nn.Sequential(nn.Linear(hidden_dim, hidden_dim), nn.LayerNorm(hidden_dim))
|
||||
self.enc_score_head = nn.Linear(hidden_dim, nc)
|
||||
self.enc_bbox_head = MLP(hidden_dim, hidden_dim, 4, num_layers=3)
|
||||
|
||||
# decoder head
|
||||
self.dec_score_head = nn.ModuleList([nn.Linear(hidden_dim, nc) for _ in range(num_decoder_layers)])
|
||||
self.dec_bbox_head = nn.ModuleList([
|
||||
MLP(hidden_dim, hidden_dim, 4, num_layers=3) for _ in range(num_decoder_layers)])
|
||||
|
||||
self._reset_parameters()
|
||||
|
||||
def forward(self, feats, gt_meta=None):
|
||||
# input projection and embedding
|
||||
memory, spatial_shapes, _ = self._get_encoder_input(feats)
|
||||
|
||||
# prepare denoising training
|
||||
if self.training:
|
||||
raise NotImplementedError
|
||||
# denoising_class, denoising_bbox_unact, attn_mask, dn_meta = \
|
||||
# get_contrastive_denoising_training_group(gt_meta,
|
||||
# self.num_classes,
|
||||
# self.num_queries,
|
||||
# self.denoising_class_embed.weight,
|
||||
# self.num_denoising,
|
||||
# self.label_noise_ratio,
|
||||
# self.box_noise_scale)
|
||||
else:
|
||||
denoising_class, denoising_bbox_unact, attn_mask = None, None, None
|
||||
|
||||
target, init_ref_points_unact, enc_topk_bboxes, enc_topk_logits = \
|
||||
self._get_decoder_input(memory, spatial_shapes, denoising_class, denoising_bbox_unact)
|
||||
|
||||
# decoder
|
||||
out_bboxes, out_logits = self.decoder(target,
|
||||
init_ref_points_unact,
|
||||
memory,
|
||||
spatial_shapes,
|
||||
self.dec_bbox_head,
|
||||
self.dec_score_head,
|
||||
self.query_pos_head,
|
||||
attn_mask=attn_mask)
|
||||
if not self.training:
|
||||
out_logits = out_logits.sigmoid_()
|
||||
return out_bboxes, out_logits # enc_topk_bboxes, enc_topk_logits, dn_meta
|
||||
|
||||
def _reset_parameters(self):
|
||||
# class and bbox head init
|
||||
bias_cls = bias_init_with_prob(0.01)
|
||||
linear_init_(self.enc_score_head)
|
||||
constant_(self.enc_score_head.bias, bias_cls)
|
||||
constant_(self.enc_bbox_head.layers[-1].weight, 0.)
|
||||
constant_(self.enc_bbox_head.layers[-1].bias, 0.)
|
||||
for cls_, reg_ in zip(self.dec_score_head, self.dec_bbox_head):
|
||||
linear_init_(cls_)
|
||||
constant_(cls_.bias, bias_cls)
|
||||
constant_(reg_.layers[-1].weight, 0.)
|
||||
constant_(reg_.layers[-1].bias, 0.)
|
||||
|
||||
linear_init_(self.enc_output[0])
|
||||
xavier_uniform_(self.enc_output[0].weight)
|
||||
if self.learnt_init_query:
|
||||
xavier_uniform_(self.tgt_embed.weight)
|
||||
xavier_uniform_(self.query_pos_head.layers[0].weight)
|
||||
xavier_uniform_(self.query_pos_head.layers[1].weight)
|
||||
for layer in self.input_proj:
|
||||
xavier_uniform_(layer[0].weight)
|
||||
|
||||
def _build_input_proj_layer(self, ch):
|
||||
self.input_proj = nn.ModuleList()
|
||||
for in_channels in ch:
|
||||
self.input_proj.append(
|
||||
nn.Sequential(nn.Conv2d(in_channels, self.hidden_dim, kernel_size=1, bias=False),
|
||||
nn.BatchNorm2d(self.hidden_dim)))
|
||||
in_channels = ch[-1]
|
||||
for _ in range(self.nl - len(ch)):
|
||||
self.input_proj.append(
|
||||
nn.Sequential(nn.Conv2D(in_channels, self.hidden_dim, kernel_size=3, stride=2, padding=1, bias=False),
|
||||
nn.BatchNorm2d(self.hidden_dim)))
|
||||
in_channels = self.hidden_dim
|
||||
|
||||
def _generate_anchors(self, spatial_shapes, grid_size=0.05, dtype=torch.float32, device='cpu', eps=1e-2):
|
||||
anchors = []
|
||||
for lvl, (h, w) in enumerate(spatial_shapes):
|
||||
grid_y, grid_x = torch.meshgrid(torch.arange(end=h, dtype=torch.float32),
|
||||
torch.arange(end=w, dtype=torch.float32),
|
||||
indexing='ij')
|
||||
grid_xy = torch.stack([grid_x, grid_y], -1)
|
||||
|
||||
valid_WH = torch.tensor([h, w]).to(torch.float32)
|
||||
grid_xy = (grid_xy.unsqueeze(0) + 0.5) / valid_WH
|
||||
wh = torch.ones_like(grid_xy) * grid_size * (2.0 ** lvl)
|
||||
anchors.append(torch.concat([grid_xy, wh], -1).reshape([-1, h * w, 4]))
|
||||
|
||||
anchors = torch.concat(anchors, 1)
|
||||
valid_mask = ((anchors > eps) * (anchors < 1 - eps)).all(-1, keepdim=True)
|
||||
anchors = torch.log(anchors / (1 - anchors))
|
||||
anchors = torch.where(valid_mask, anchors, torch.inf)
|
||||
return anchors.to(device=device, dtype=dtype), valid_mask.to(device=device)
|
||||
|
||||
def _get_encoder_input(self, feats):
|
||||
# get projection features
|
||||
proj_feats = [self.input_proj[i](feat) for i, feat in enumerate(feats)]
|
||||
if self.nl > len(proj_feats):
|
||||
len_srcs = len(proj_feats)
|
||||
for i in range(len_srcs, self.nl):
|
||||
if i == len_srcs:
|
||||
proj_feats.append(self.input_proj[i](feats[-1]))
|
||||
else:
|
||||
proj_feats.append(self.input_proj[i](proj_feats[-1]))
|
||||
|
||||
# get encoder inputs
|
||||
feat_flatten = []
|
||||
spatial_shapes = []
|
||||
level_start_index = [0]
|
||||
for feat in proj_feats:
|
||||
_, _, h, w = feat.shape
|
||||
# [b, c, h, w] -> [b, h*w, c]
|
||||
feat_flatten.append(feat.flatten(2).permute(0, 2, 1))
|
||||
# [nl, 2]
|
||||
spatial_shapes.append([h, w])
|
||||
# [l], start index of each level
|
||||
level_start_index.append(h * w + level_start_index[-1])
|
||||
|
||||
# [b, l, c]
|
||||
feat_flatten = torch.concat(feat_flatten, 1)
|
||||
level_start_index.pop()
|
||||
return feat_flatten, spatial_shapes, level_start_index
|
||||
|
||||
def _get_decoder_input(self, memory, spatial_shapes, denoising_class=None, denoising_bbox_unact=None):
|
||||
bs, _, _ = memory.shape
|
||||
# prepare input for decoder
|
||||
anchors, valid_mask = self._generate_anchors(spatial_shapes, dtype=memory.dtype, device=memory.device)
|
||||
memory = torch.where(valid_mask, memory, torch.tensor(0.))
|
||||
output_memory = self.enc_output(memory)
|
||||
|
||||
enc_outputs_class = self.enc_score_head(output_memory) # (bs, h*w, nc)
|
||||
enc_outputs_coord_unact = self.enc_bbox_head(output_memory) + anchors # (bs, h*w, 4)
|
||||
|
||||
# (bs, topk)
|
||||
_, topk_ind = torch.topk(enc_outputs_class.max(-1).values, self.num_queries, dim=1)
|
||||
# extract region proposal boxes
|
||||
# (bs, topk_ind)
|
||||
batch_ind = torch.arange(end=bs, dtype=topk_ind.dtype).unsqueeze(-1).repeat(1, self.num_queries).view(-1)
|
||||
topk_ind = topk_ind.view(-1)
|
||||
|
||||
# Unsigmoided
|
||||
reference_points_unact = enc_outputs_coord_unact[batch_ind, topk_ind].view(bs, self.num_queries, -1)
|
||||
|
||||
enc_topk_bboxes = torch.sigmoid(reference_points_unact)
|
||||
if denoising_bbox_unact is not None:
|
||||
reference_points_unact = torch.concat([denoising_bbox_unact, reference_points_unact], 1)
|
||||
if self.training:
|
||||
reference_points_unact = reference_points_unact.detach()
|
||||
enc_topk_logits = enc_outputs_class[batch_ind, topk_ind].view(bs, self.num_queries, -1)
|
||||
|
||||
# extract region features
|
||||
if self.learnt_init_query:
|
||||
target = self.tgt_embed.weight.unsqueeze(0).repeat(bs, 1, 1)
|
||||
else:
|
||||
target = output_memory[batch_ind, topk_ind].view(bs, self.num_queries, -1)
|
||||
if self.training:
|
||||
target = target.detach()
|
||||
if denoising_class is not None:
|
||||
target = torch.concat([denoising_class, target], 1)
|
||||
|
||||
return target, reference_points_unact, enc_topk_bboxes, enc_topk_logits
|
||||
390
ultralytics/nn/modules/transformer.py
Normal file
390
ultralytics/nn/modules/transformer.py
Normal file
@ -0,0 +1,390 @@
|
||||
# Ultralytics YOLO 🚀, AGPL-3.0 license
|
||||
"""
|
||||
Transformer modules
|
||||
"""
|
||||
|
||||
import math
|
||||
|
||||
import torch
|
||||
import torch.nn as nn
|
||||
import torch.nn.functional as F
|
||||
from torch.nn.init import constant_, xavier_uniform_
|
||||
|
||||
from .conv import Conv
|
||||
from .utils import _get_clones, inverse_sigmoid, multi_scale_deformable_attn_pytorch
|
||||
|
||||
__all__ = [
|
||||
'TransformerEncoderLayer', 'TransformerLayer', 'TransformerBlock', 'MLPBlock', 'LayerNorm2d', 'AIFI',
|
||||
'DeformableTransformerDecoder', 'DeformableTransformerDecoderLayer', 'MSDeformAttn', 'MLP']
|
||||
|
||||
|
||||
class TransformerEncoderLayer(nn.Module):
|
||||
"""Transformer Encoder."""
|
||||
|
||||
def __init__(self, c1, cm=2048, num_heads=8, dropout=0.0, act=nn.GELU(), normalize_before=False):
|
||||
super().__init__()
|
||||
self.ma = nn.MultiheadAttention(c1, num_heads, dropout=dropout, batch_first=True)
|
||||
# Implementation of Feedforward model
|
||||
self.fc1 = nn.Linear(c1, cm)
|
||||
self.fc2 = nn.Linear(cm, c1)
|
||||
|
||||
self.norm1 = nn.LayerNorm(c1)
|
||||
self.norm2 = nn.LayerNorm(c1)
|
||||
self.dropout = nn.Dropout(dropout)
|
||||
self.dropout1 = nn.Dropout(dropout)
|
||||
self.dropout2 = nn.Dropout(dropout)
|
||||
|
||||
self.act = act
|
||||
self.normalize_before = normalize_before
|
||||
|
||||
def with_pos_embed(self, tensor, pos=None):
|
||||
"""Add position embeddings if given."""
|
||||
return tensor if pos is None else tensor + pos
|
||||
|
||||
def forward_post(self, src, src_mask=None, src_key_padding_mask=None, pos=None):
|
||||
q = k = self.with_pos_embed(src, pos)
|
||||
src2 = self.ma(q, k, value=src, attn_mask=src_mask, key_padding_mask=src_key_padding_mask)[0]
|
||||
src = src + self.dropout1(src2)
|
||||
src = self.norm1(src)
|
||||
src2 = self.fc2(self.dropout(self.act(self.fc1(src))))
|
||||
src = src + self.dropout2(src2)
|
||||
src = self.norm2(src)
|
||||
return src
|
||||
|
||||
def forward_pre(self, src, src_mask=None, src_key_padding_mask=None, pos=None):
|
||||
src2 = self.norm1(src)
|
||||
q = k = self.with_pos_embed(src2, pos)
|
||||
src2 = self.ma(q, k, value=src2, attn_mask=src_mask, key_padding_mask=src_key_padding_mask)[0]
|
||||
src = src + self.dropout1(src2)
|
||||
src2 = self.norm2(src)
|
||||
src2 = self.fc2(self.dropout(self.act(self.fc1(src2))))
|
||||
src = src + self.dropout2(src2)
|
||||
return src
|
||||
|
||||
def forward(self, src, src_mask=None, src_key_padding_mask=None, pos=None):
|
||||
"""Forward propagates the input through the encoder module."""
|
||||
if self.normalize_before:
|
||||
return self.forward_pre(src, src_mask, src_key_padding_mask, pos)
|
||||
return self.forward_post(src, src_mask, src_key_padding_mask, pos)
|
||||
|
||||
|
||||
class AIFI(TransformerEncoderLayer):
|
||||
|
||||
def __init__(self, c1, cm=2048, num_heads=8, dropout=0, act=nn.GELU(), normalize_before=False):
|
||||
super().__init__(c1, cm, num_heads, dropout, act, normalize_before)
|
||||
|
||||
def forward(self, x):
|
||||
c, h, w = x.shape[1:]
|
||||
pos_embed = self.build_2d_sincos_position_embedding(w, h, c)
|
||||
# flatten [B, C, H, W] to [B, HxW, C]
|
||||
x = super().forward(x.flatten(2).permute(0, 2, 1), pos=pos_embed.to(device=x.device, dtype=x.dtype))
|
||||
return x.permute((0, 2, 1)).view([-1, c, h, w])
|
||||
|
||||
@staticmethod
|
||||
def build_2d_sincos_position_embedding(w, h, embed_dim=256, temperature=10000.):
|
||||
grid_w = torch.arange(int(w), dtype=torch.float32)
|
||||
grid_h = torch.arange(int(h), dtype=torch.float32)
|
||||
grid_w, grid_h = torch.meshgrid(grid_w, grid_h, indexing='ij')
|
||||
assert embed_dim % 4 == 0, \
|
||||
'Embed dimension must be divisible by 4 for 2D sin-cos position embedding'
|
||||
pos_dim = embed_dim // 4
|
||||
omega = torch.arange(pos_dim, dtype=torch.float32) / pos_dim
|
||||
omega = 1. / (temperature ** omega)
|
||||
|
||||
out_w = grid_w.flatten()[..., None] @ omega[None]
|
||||
out_h = grid_h.flatten()[..., None] @ omega[None]
|
||||
|
||||
return torch.concat([torch.sin(out_w), torch.cos(out_w),
|
||||
torch.sin(out_h), torch.cos(out_h)], axis=1)[None, :, :]
|
||||
|
||||
|
||||
class TransformerLayer(nn.Module):
|
||||
"""Transformer layer https://arxiv.org/abs/2010.11929 (LayerNorm layers removed for better performance)."""
|
||||
|
||||
def __init__(self, c, num_heads):
|
||||
"""Initializes a self-attention mechanism using linear transformations and multi-head attention."""
|
||||
super().__init__()
|
||||
self.q = nn.Linear(c, c, bias=False)
|
||||
self.k = nn.Linear(c, c, bias=False)
|
||||
self.v = nn.Linear(c, c, bias=False)
|
||||
self.ma = nn.MultiheadAttention(embed_dim=c, num_heads=num_heads)
|
||||
self.fc1 = nn.Linear(c, c, bias=False)
|
||||
self.fc2 = nn.Linear(c, c, bias=False)
|
||||
|
||||
def forward(self, x):
|
||||
"""Apply a transformer block to the input x and return the output."""
|
||||
x = self.ma(self.q(x), self.k(x), self.v(x))[0] + x
|
||||
x = self.fc2(self.fc1(x)) + x
|
||||
return x
|
||||
|
||||
|
||||
class TransformerBlock(nn.Module):
|
||||
"""Vision Transformer https://arxiv.org/abs/2010.11929."""
|
||||
|
||||
def __init__(self, c1, c2, num_heads, num_layers):
|
||||
"""Initialize a Transformer module with position embedding and specified number of heads and layers."""
|
||||
super().__init__()
|
||||
self.conv = None
|
||||
if c1 != c2:
|
||||
self.conv = Conv(c1, c2)
|
||||
self.linear = nn.Linear(c2, c2) # learnable position embedding
|
||||
self.tr = nn.Sequential(*(TransformerLayer(c2, num_heads) for _ in range(num_layers)))
|
||||
self.c2 = c2
|
||||
|
||||
def forward(self, x):
|
||||
"""Forward propagates the input through the bottleneck module."""
|
||||
if self.conv is not None:
|
||||
x = self.conv(x)
|
||||
b, _, w, h = x.shape
|
||||
p = x.flatten(2).permute(2, 0, 1)
|
||||
return self.tr(p + self.linear(p)).permute(1, 2, 0).reshape(b, self.c2, w, h)
|
||||
|
||||
|
||||
class MLPBlock(nn.Module):
|
||||
|
||||
def __init__(self, embedding_dim, mlp_dim, act=nn.GELU):
|
||||
super().__init__()
|
||||
self.lin1 = nn.Linear(embedding_dim, mlp_dim)
|
||||
self.lin2 = nn.Linear(mlp_dim, embedding_dim)
|
||||
self.act = act()
|
||||
|
||||
def forward(self, x: torch.Tensor) -> torch.Tensor:
|
||||
return self.lin2(self.act(self.lin1(x)))
|
||||
|
||||
|
||||
class MLP(nn.Module):
|
||||
""" Very simple multi-layer perceptron (also called FFN)"""
|
||||
|
||||
def __init__(self, input_dim, hidden_dim, output_dim, num_layers):
|
||||
super().__init__()
|
||||
self.num_layers = num_layers
|
||||
h = [hidden_dim] * (num_layers - 1)
|
||||
self.layers = nn.ModuleList(nn.Linear(n, k) for n, k in zip([input_dim] + h, h + [output_dim]))
|
||||
|
||||
def forward(self, x):
|
||||
for i, layer in enumerate(self.layers):
|
||||
x = F.relu(layer(x)) if i < self.num_layers - 1 else layer(x)
|
||||
return x
|
||||
|
||||
|
||||
# From https://github.com/facebookresearch/detectron2/blob/main/detectron2/layers/batch_norm.py # noqa
|
||||
# Itself from https://github.com/facebookresearch/ConvNeXt/blob/d1fa8f6fef0a165b27399986cc2bdacc92777e40/models/convnext.py#L119 # noqa
|
||||
class LayerNorm2d(nn.Module):
|
||||
|
||||
def __init__(self, num_channels, eps=1e-6):
|
||||
super().__init__()
|
||||
self.weight = nn.Parameter(torch.ones(num_channels))
|
||||
self.bias = nn.Parameter(torch.zeros(num_channels))
|
||||
self.eps = eps
|
||||
|
||||
def forward(self, x):
|
||||
u = x.mean(1, keepdim=True)
|
||||
s = (x - u).pow(2).mean(1, keepdim=True)
|
||||
x = (x - u) / torch.sqrt(s + self.eps)
|
||||
x = self.weight[:, None, None] * x + self.bias[:, None, None]
|
||||
return x
|
||||
|
||||
|
||||
class MSDeformAttn(nn.Module):
|
||||
"""
|
||||
Original Multi-Scale Deformable Attention Module.
|
||||
https://github.com/fundamentalvision/Deformable-DETR/blob/main/models/ops/modules/ms_deform_attn.py
|
||||
"""
|
||||
|
||||
def __init__(self, d_model=256, n_levels=4, n_heads=8, n_points=4):
|
||||
super().__init__()
|
||||
if d_model % n_heads != 0:
|
||||
raise ValueError(f'd_model must be divisible by n_heads, but got {d_model} and {n_heads}')
|
||||
_d_per_head = d_model // n_heads
|
||||
# you'd better set _d_per_head to a power of 2 which is more efficient in our CUDA implementation
|
||||
assert _d_per_head * n_heads == d_model, '`d_model` must be divisible by `n_heads`'
|
||||
|
||||
self.im2col_step = 64
|
||||
|
||||
self.d_model = d_model
|
||||
self.n_levels = n_levels
|
||||
self.n_heads = n_heads
|
||||
self.n_points = n_points
|
||||
|
||||
self.sampling_offsets = nn.Linear(d_model, n_heads * n_levels * n_points * 2)
|
||||
self.attention_weights = nn.Linear(d_model, n_heads * n_levels * n_points)
|
||||
self.value_proj = nn.Linear(d_model, d_model)
|
||||
self.output_proj = nn.Linear(d_model, d_model)
|
||||
|
||||
self._reset_parameters()
|
||||
|
||||
def _reset_parameters(self):
|
||||
constant_(self.sampling_offsets.weight.data, 0.)
|
||||
thetas = torch.arange(self.n_heads, dtype=torch.float32) * (2.0 * math.pi / self.n_heads)
|
||||
grid_init = torch.stack([thetas.cos(), thetas.sin()], -1)
|
||||
grid_init = (grid_init / grid_init.abs().max(-1, keepdim=True)[0]).view(self.n_heads, 1, 1, 2).repeat(
|
||||
1, self.n_levels, self.n_points, 1)
|
||||
for i in range(self.n_points):
|
||||
grid_init[:, :, i, :] *= i + 1
|
||||
with torch.no_grad():
|
||||
self.sampling_offsets.bias = nn.Parameter(grid_init.view(-1))
|
||||
constant_(self.attention_weights.weight.data, 0.)
|
||||
constant_(self.attention_weights.bias.data, 0.)
|
||||
xavier_uniform_(self.value_proj.weight.data)
|
||||
constant_(self.value_proj.bias.data, 0.)
|
||||
xavier_uniform_(self.output_proj.weight.data)
|
||||
constant_(self.output_proj.bias.data, 0.)
|
||||
|
||||
def forward(self, query, reference_points, value, value_spatial_shapes, value_mask=None):
|
||||
"""
|
||||
https://github.com/PaddlePaddle/PaddleDetection/blob/develop/ppdet/modeling/transformers/deformable_transformer.py
|
||||
Args:
|
||||
query (Tensor): [bs, query_length, C]
|
||||
reference_points (Tensor): [bs, query_length, n_levels, 2], range in [0, 1], top-left (0,0),
|
||||
bottom-right (1, 1), including padding area
|
||||
value (Tensor): [bs, value_length, C]
|
||||
value_spatial_shapes (List): [n_levels, 2], [(H_0, W_0), (H_1, W_1), ..., (H_{L-1}, W_{L-1})]
|
||||
value_mask (Tensor): [bs, value_length], True for non-padding elements, False for padding elements
|
||||
|
||||
Returns:
|
||||
output (Tensor): [bs, Length_{query}, C]
|
||||
"""
|
||||
bs, len_q = query.shape[:2]
|
||||
_, len_v = value.shape[:2]
|
||||
assert sum(s[0] * s[1] for s in value_spatial_shapes) == len_v
|
||||
|
||||
value = self.value_proj(value)
|
||||
if value_mask is not None:
|
||||
value = value.masked_fill(value_mask[..., None], float(0))
|
||||
value = value.view(bs, len_v, self.n_heads, self.d_model // self.n_heads)
|
||||
sampling_offsets = self.sampling_offsets(query).view(bs, len_q, self.n_heads, self.n_levels, self.n_points, 2)
|
||||
attention_weights = self.attention_weights(query).view(bs, len_q, self.n_heads, self.n_levels * self.n_points)
|
||||
attention_weights = F.softmax(attention_weights, -1).view(bs, len_q, self.n_heads, self.n_levels, self.n_points)
|
||||
# N, Len_q, n_heads, n_levels, n_points, 2
|
||||
n = reference_points.shape[-1]
|
||||
if n == 2:
|
||||
offset_normalizer = torch.as_tensor(value_spatial_shapes, dtype=query.dtype, device=query.device).flip(-1)
|
||||
add = sampling_offsets / offset_normalizer[None, None, None, :, None, :]
|
||||
sampling_locations = reference_points[:, :, None, :, None, :] + add
|
||||
|
||||
elif n == 4:
|
||||
add = sampling_offsets / self.n_points * reference_points[:, :, None, :, None, 2:] * 0.5
|
||||
sampling_locations = reference_points[:, :, None, :, None, :2] + add
|
||||
else:
|
||||
raise ValueError(f'Last dim of reference_points must be 2 or 4, but got {n}.')
|
||||
output = multi_scale_deformable_attn_pytorch(value, value_spatial_shapes, sampling_locations, attention_weights)
|
||||
output = self.output_proj(output)
|
||||
return output
|
||||
|
||||
|
||||
class DeformableTransformerDecoderLayer(nn.Module):
|
||||
"""
|
||||
https://github.com/PaddlePaddle/PaddleDetection/blob/develop/ppdet/modeling/transformers/deformable_transformer.py
|
||||
https://github.com/fundamentalvision/Deformable-DETR/blob/main/models/deformable_transformer.py
|
||||
"""
|
||||
|
||||
def __init__(self, d_model=256, n_heads=8, d_ffn=1024, dropout=0., act=nn.ReLU(), n_levels=4, n_points=4):
|
||||
super().__init__()
|
||||
|
||||
# self attention
|
||||
self.self_attn = nn.MultiheadAttention(d_model, n_heads, dropout=dropout)
|
||||
self.dropout1 = nn.Dropout(dropout)
|
||||
self.norm1 = nn.LayerNorm(d_model)
|
||||
|
||||
# cross attention
|
||||
self.cross_attn = MSDeformAttn(d_model, n_levels, n_heads, n_points)
|
||||
self.dropout2 = nn.Dropout(dropout)
|
||||
self.norm2 = nn.LayerNorm(d_model)
|
||||
|
||||
# ffn
|
||||
self.linear1 = nn.Linear(d_model, d_ffn)
|
||||
self.act = act
|
||||
self.dropout3 = nn.Dropout(dropout)
|
||||
self.linear2 = nn.Linear(d_ffn, d_model)
|
||||
self.dropout4 = nn.Dropout(dropout)
|
||||
self.norm3 = nn.LayerNorm(d_model)
|
||||
|
||||
@staticmethod
|
||||
def with_pos_embed(tensor, pos):
|
||||
return tensor if pos is None else tensor + pos
|
||||
|
||||
def forward_ffn(self, tgt):
|
||||
tgt2 = self.linear2(self.dropout3(self.act(self.linear1(tgt))))
|
||||
tgt = tgt + self.dropout4(tgt2)
|
||||
tgt = self.norm3(tgt)
|
||||
return tgt
|
||||
|
||||
def forward(self,
|
||||
tgt,
|
||||
reference_points,
|
||||
src,
|
||||
src_spatial_shapes,
|
||||
src_padding_mask=None,
|
||||
attn_mask=None,
|
||||
query_pos=None):
|
||||
# self attention
|
||||
q = k = self.with_pos_embed(tgt, query_pos)
|
||||
if attn_mask is not None:
|
||||
attn_mask = torch.where(attn_mask.astype('bool'), torch.zeros(attn_mask.shape, tgt.dtype),
|
||||
torch.full(attn_mask.shape, float('-inf'), tgt.dtype))
|
||||
tgt2 = self.self_attn(q.transpose(0, 1), k.transpose(0, 1), tgt.transpose(0, 1))[0].transpose(0, 1)
|
||||
tgt = tgt + self.dropout1(tgt2)
|
||||
tgt = self.norm1(tgt)
|
||||
|
||||
# cross attention
|
||||
tgt2 = self.cross_attn(self.with_pos_embed(tgt, query_pos), reference_points, src, src_spatial_shapes,
|
||||
src_padding_mask)
|
||||
tgt = tgt + self.dropout2(tgt2)
|
||||
tgt = self.norm2(tgt)
|
||||
|
||||
# ffn
|
||||
tgt = self.forward_ffn(tgt)
|
||||
|
||||
return tgt
|
||||
|
||||
|
||||
class DeformableTransformerDecoder(nn.Module):
|
||||
"""
|
||||
https://github.com/PaddlePaddle/PaddleDetection/blob/develop/ppdet/modeling/transformers/deformable_transformer.py
|
||||
"""
|
||||
|
||||
def __init__(self, hidden_dim, decoder_layer, num_layers, eval_idx=-1):
|
||||
super().__init__()
|
||||
self.layers = _get_clones(decoder_layer, num_layers)
|
||||
self.num_layers = num_layers
|
||||
self.hidden_dim = hidden_dim
|
||||
self.eval_idx = eval_idx if eval_idx >= 0 else num_layers + eval_idx
|
||||
|
||||
def forward(self,
|
||||
tgt,
|
||||
reference_points,
|
||||
src,
|
||||
src_spatial_shapes,
|
||||
bbox_head,
|
||||
score_head,
|
||||
query_pos_head,
|
||||
attn_mask=None,
|
||||
src_padding_mask=None):
|
||||
output = tgt
|
||||
dec_out_bboxes = []
|
||||
dec_out_logits = []
|
||||
ref_points = None
|
||||
ref_points_detach = torch.sigmoid(reference_points)
|
||||
for i, layer in enumerate(self.layers):
|
||||
ref_points_input = ref_points_detach.unsqueeze(2)
|
||||
query_pos_embed = query_pos_head(ref_points_detach)
|
||||
output = layer(output, ref_points_input, src, src_spatial_shapes, src_padding_mask, attn_mask,
|
||||
query_pos_embed)
|
||||
|
||||
inter_ref_bbox = torch.sigmoid(bbox_head[i](output) + inverse_sigmoid(ref_points_detach))
|
||||
|
||||
if self.training:
|
||||
dec_out_logits.append(score_head[i](output))
|
||||
if i == 0:
|
||||
dec_out_bboxes.append(inter_ref_bbox)
|
||||
else:
|
||||
dec_out_bboxes.append(torch.sigmoid(bbox_head[i](output) + inverse_sigmoid(ref_points)))
|
||||
elif i == self.eval_idx:
|
||||
dec_out_logits.append(score_head[i](output))
|
||||
dec_out_bboxes.append(inter_ref_bbox)
|
||||
break
|
||||
|
||||
ref_points = inter_ref_bbox
|
||||
ref_points_detach = inter_ref_bbox.detach() if self.training else inter_ref_bbox
|
||||
|
||||
return torch.stack(dec_out_bboxes), torch.stack(dec_out_logits)
|
||||
78
ultralytics/nn/modules/utils.py
Normal file
78
ultralytics/nn/modules/utils.py
Normal file
@ -0,0 +1,78 @@
|
||||
# Ultralytics YOLO 🚀, AGPL-3.0 license
|
||||
"""
|
||||
Module utils
|
||||
"""
|
||||
|
||||
import copy
|
||||
import math
|
||||
|
||||
import numpy as np
|
||||
import torch
|
||||
import torch.nn as nn
|
||||
import torch.nn.functional as F
|
||||
from torch.nn.init import uniform_
|
||||
|
||||
__all__ = ['multi_scale_deformable_attn_pytorch', 'inverse_sigmoid']
|
||||
|
||||
|
||||
def _get_clones(module, n):
|
||||
return nn.ModuleList([copy.deepcopy(module) for _ in range(n)])
|
||||
|
||||
|
||||
def bias_init_with_prob(prior_prob=0.01):
|
||||
"""initialize conv/fc bias value according to a given probability value."""
|
||||
return float(-np.log((1 - prior_prob) / prior_prob)) # return bias_init
|
||||
|
||||
|
||||
def linear_init_(module):
|
||||
bound = 1 / math.sqrt(module.weight.shape[0])
|
||||
uniform_(module.weight, -bound, bound)
|
||||
if hasattr(module, 'bias') and module.bias is not None:
|
||||
uniform_(module.bias, -bound, bound)
|
||||
|
||||
|
||||
def inverse_sigmoid(x, eps=1e-5):
|
||||
x = x.clamp(min=0, max=1)
|
||||
x1 = x.clamp(min=eps)
|
||||
x2 = (1 - x).clamp(min=eps)
|
||||
return torch.log(x1 / x2)
|
||||
|
||||
|
||||
def multi_scale_deformable_attn_pytorch(value: torch.Tensor, value_spatial_shapes: torch.Tensor,
|
||||
sampling_locations: torch.Tensor,
|
||||
attention_weights: torch.Tensor) -> torch.Tensor:
|
||||
"""
|
||||
Multi-scale deformable attention.
|
||||
https://github.com/IDEA-Research/detrex/blob/main/detrex/layers/multi_scale_deform_attn.py
|
||||
"""
|
||||
|
||||
bs, _, num_heads, embed_dims = value.shape
|
||||
_, num_queries, num_heads, num_levels, num_points, _ = sampling_locations.shape
|
||||
value_list = value.split([H_ * W_ for H_, W_ in value_spatial_shapes], dim=1)
|
||||
sampling_grids = 2 * sampling_locations - 1
|
||||
sampling_value_list = []
|
||||
for level, (H_, W_) in enumerate(value_spatial_shapes):
|
||||
# bs, H_*W_, num_heads, embed_dims ->
|
||||
# bs, H_*W_, num_heads*embed_dims ->
|
||||
# bs, num_heads*embed_dims, H_*W_ ->
|
||||
# bs*num_heads, embed_dims, H_, W_
|
||||
value_l_ = (value_list[level].flatten(2).transpose(1, 2).reshape(bs * num_heads, embed_dims, H_, W_))
|
||||
# bs, num_queries, num_heads, num_points, 2 ->
|
||||
# bs, num_heads, num_queries, num_points, 2 ->
|
||||
# bs*num_heads, num_queries, num_points, 2
|
||||
sampling_grid_l_ = sampling_grids[:, :, :, level].transpose(1, 2).flatten(0, 1)
|
||||
# bs*num_heads, embed_dims, num_queries, num_points
|
||||
sampling_value_l_ = F.grid_sample(value_l_,
|
||||
sampling_grid_l_,
|
||||
mode='bilinear',
|
||||
padding_mode='zeros',
|
||||
align_corners=False)
|
||||
sampling_value_list.append(sampling_value_l_)
|
||||
# (bs, num_queries, num_heads, num_levels, num_points) ->
|
||||
# (bs, num_heads, num_queries, num_levels, num_points) ->
|
||||
# (bs, num_heads, 1, num_queries, num_levels*num_points)
|
||||
attention_weights = attention_weights.transpose(1, 2).reshape(bs * num_heads, 1, num_queries,
|
||||
num_levels * num_points)
|
||||
output = ((torch.stack(sampling_value_list, dim=-2).flatten(-2) * attention_weights).sum(-1).view(
|
||||
bs, num_heads * embed_dims, num_queries))
|
||||
return output.transpose(1, 2).contiguous()
|
||||
Reference in New Issue
Block a user