YOLOv7 网络结构图详解
- yolo.py 输出结构
- 整体图
- yolov7.yaml
- 组件结构
- CBS 模块
- ELAN1
- ELAN2
- MP1&2
- MP1
- MP2
- SPPCSPC
- 参考
yolo.py 输出结构
输出的 arguments 和yaml文件的区别就是 多了第一列Conv输入的通道数
python">YOLOR v0.1-112-g55b90e1 torch 1.7.0 CUDA:0 (Quadro RTX 4000, 8191.6875MB)
from n params module arguments
0 -1 1 928 models.common.Conv [3, 32, 3, 1]
1 -1 1 18560 models.common.Conv [32, 64, 3, 2]
2 -1 1 36992 models.common.Conv [64, 64, 3, 1]
3 -1 1 73984 models.common.Conv [64, 128, 3, 2]
4 -1 1 8320 models.common.Conv [128, 64, 1, 1]
5 -2 1 8320 models.common.Conv [128, 64, 1, 1]
6 -1 1 36992 models.common.Conv [64, 64, 3, 1]
7 -1 1 36992 models.common.Conv [64, 64, 3, 1]
8 -1 1 36992 models.common.Conv [64, 64, 3, 1]
9 -1 1 36992 models.common.Conv [64, 64, 3, 1]
10 [-1, -3, -5, -6] 1 0 models.common.Concat [1]
11 -1 1 66048 models.common.Conv [256, 256, 1, 1]
12 -1 1 0 models.common.MP []
13 -1 1 33024 models.common.Conv [256, 128, 1, 1]
14 -3 1 33024 models.common.Conv [256, 128, 1, 1]
15 -1 1 147712 models.common.Conv [128, 128, 3, 2]
16 [-1, -3] 1 0 models.common.Concat [1]
17 -1 1 33024 models.common.Conv [256, 128, 1, 1]
18 -2 1 33024 models.common.Conv [256, 128, 1, 1]
19 -1 1 147712 models.common.Conv [128, 128, 3, 1]
20 -1 1 147712 models.common.Conv [128, 128, 3, 1]
21 -1 1 147712 models.common.Conv [128, 128, 3, 1]
22 -1 1 147712 models.common.Conv [128, 128, 3, 1]
23 [-1, -3, -5, -6] 1 0 models.common.Concat [1]
24 -1 1 263168 models.common.Conv [512, 512, 1, 1]
25 -1 1 0 models.common.MP []
26 -1 1 131584 models.common.Conv [512, 256, 1, 1]
27 -3 1 131584 models.common.Conv [512, 256, 1, 1]
28 -1 1 590336 models.common.Conv [256, 256, 3, 2]
29 [-1, -3] 1 0 models.common.Concat [1]
30 -1 1 131584 models.common.Conv [512, 256, 1, 1]
31 -2 1 131584 models.common.Conv [512, 256, 1, 1]
32 -1 1 590336 models.common.Conv [256, 256, 3, 1]
33 -1 1 590336 models.common.Conv [256, 256, 3, 1]
34 -1 1 590336 models.common.Conv [256, 256, 3, 1]
35 -1 1 590336 models.common.Conv [256, 256, 3, 1]
36 [-1, -3, -5, -6] 1 0 models.common.Concat [1]
37 -1 1 1050624 models.common.Conv [1024, 1024, 1, 1]
38 -1 1 0 models.common.MP []
39 -1 1 525312 models.common.Conv [1024, 512, 1, 1]
40 -3 1 525312 models.common.Conv [1024, 512, 1, 1]
41 -1 1 2360320 models.common.Conv [512, 512, 3, 2]
42 [-1, -3] 1 0 models.common.Concat [1]
43 -1 1 262656 models.common.Conv [1024, 256, 1, 1]
44 -2 1 262656 models.common.Conv [1024, 256, 1, 1]
45 -1 1 590336 models.common.Conv [256, 256, 3, 1]
46 -1 1 590336 models.common.Conv [256, 256, 3, 1]
47 -1 1 590336 models.common.Conv [256, 256, 3, 1]
48 -1 1 590336 models.common.Conv [256, 256, 3, 1]
49 [-1, -3, -5, -6] 1 0 models.common.Concat [1]
50 -1 1 1050624 models.common.Conv [1024, 1024, 1, 1]
51 -1 1 7609344 models.common.SPPCSPC [1024, 512, 1]
52 -1 1 131584 models.common.Conv [512, 256, 1, 1]
53 -1 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
54 37 1 262656 models.common.Conv [1024, 256, 1, 1]
55 [-1, -2] 1 0 models.common.Concat [1]
56 -1 1 131584 models.common.Conv [512, 256, 1, 1]
57 -2 1 131584 models.common.Conv [512, 256, 1, 1]
58 -1 1 295168 models.common.Conv [256, 128, 3, 1]
59 -1 1 147712 models.common.Conv [128, 128, 3, 1]
60 -1 1 147712 models.common.Conv [128, 128, 3, 1]
61 -1 1 147712 models.common.Conv [128, 128, 3, 1]
62[-1, -2, -3, -4, -5, -6] 1 0 models.common.Concat [1]
63 -1 1 262656 models.common.Conv [1024, 256, 1, 1]
64 -1 1 33024 models.common.Conv [256, 128, 1, 1]
65 -1 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
66 24 1 65792 models.common.Conv [512, 128, 1, 1]
67 [-1, -2] 1 0 models.common.Concat [1]
68 -1 1 33024 models.common.Conv [256, 128, 1, 1]
69 -2 1 33024 models.common.Conv [256, 128, 1, 1]
70 -1 1 73856 models.common.Conv [128, 64, 3, 1]
71 -1 1 36992 models.common.Conv [64, 64, 3, 1]
72 -1 1 36992 models.common.Conv [64, 64, 3, 1]
73 -1 1 36992 models.common.Conv [64, 64, 3, 1]
74[-1, -2, -3, -4, -5, -6] 1 0 models.common.Concat [1]
75 -1 1 65792 models.common.Conv [512, 128, 1, 1]
76 -1 1 0 models.common.MP []
77 -1 1 16640 models.common.Conv [128, 128, 1, 1]
78 -3 1 16640 models.common.Conv [128, 128, 1, 1]
79 -1 1 147712 models.common.Conv [128, 128, 3, 2]
80 [-1, -3, 63] 1 0 models.common.Concat [1]
81 -1 1 131584 models.common.Conv [512, 256, 1, 1]
82 -2 1 131584 models.common.Conv [512, 256, 1, 1]
83 -1 1 295168 models.common.Conv [256, 128, 3, 1]
84 -1 1 147712 models.common.Conv [128, 128, 3, 1]
85 -1 1 147712 models.common.Conv [128, 128, 3, 1]
86 -1 1 147712 models.common.Conv [128, 128, 3, 1]
87[-1, -2, -3, -4, -5, -6] 1 0 models.common.Concat [1]
88 -1 1 262656 models.common.Conv [1024, 256, 1, 1]
89 -1 1 0 models.common.MP []
90 -1 1 66048 models.common.Conv [256, 256, 1, 1]
91 -3 1 66048 models.common.Conv [256, 256, 1, 1]
92 -1 1 590336 models.common.Conv [256, 256, 3, 2]
93 [-1, -3, 51] 1 0 models.common.Concat [1]
94 -1 1 525312 models.common.Conv [1024, 512, 1, 1]
95 -2 1 525312 models.common.Conv [1024, 512, 1, 1]
96 -1 1 1180160 models.common.Conv [512, 256, 3, 1]
97 -1 1 590336 models.common.Conv [256, 256, 3, 1]
98 -1 1 590336 models.common.Conv [256, 256, 3, 1]
99 -1 1 590336 models.common.Conv [256, 256, 3, 1]
100[-1, -2, -3, -4, -5, -6] 1 0 models.common.Concat [1]
101 -1 1 1049600 models.common.Conv [2048, 512, 1, 1]
102 75 1 328704 models.common.RepConv [128, 256, 3, 1]
103 88 1 1312768 models.common.RepConv [256, 512, 3, 1]
104 101 1 5246976 models.common.RepConv [512, 1024, 3, 1]
105 [102, 103, 104] 1 39550 IDetect [2, [[12, 16, 19, 36, 40, 28], [36, 75, 76, 55, 72, 146], [142, 110, 192, 243, 459, 401]], [256, 512, 1024]]
Model Summary: 415 layers, 37201950 parameters, 37201950 gradients, 105.1 GFLOPS
整体图
整体图如下所示，这个有有yaml层数，下一张有具体输出，第三张b导的简洁一些，结合3张图起来看配合yaml文件，基本就很好理解了。
yolov7.yaml
[-1, 1, Conv, [32, 3, 1] 其中的[32, 3, 1] 表示输出通道数为32 ，卷积核为3*3，步长为2
边看整体网络结构图，边看yaml文件，对着看。
注意：
backbone 和 head中的模块MP-1和MP-2区别，backbone中尺寸减半通道数不变，head中尺寸减半通道数变成两倍
backbone 和 head中的模块ELAN-1和ELAN-2的区别，banbone中通道数变成两倍，head中减半
ELAN在backbone中扩张我估计是为了更好的提取特征，而MP-1通道数减半，可以把它理解为改进版本的下采样。
python"># parameters
nc: 2 # number of classes
depth_multiple: 1.0 # model depth multiple
width_multiple: 1.0 # layer channel multiple
# anchors
anchors:
- [12,16, 19,36, 40,28] # P3/8
- [36,75, 76,55, 72,146] # P4/16
- [142,110, 192,243, 459,401] # P5/32
# yolov7 backbone
backbone:
# [from, number, module, args] 640*640*3
[[-1, 1, Conv, [32, 3, 1]], # 0 640*640*32
[-1, 1, Conv, [64, 3, 2]], # 1-P1/2 320*320*64
[-1, 1, Conv, [64, 3, 1]], # 320*320*64
[-1, 1, Conv, [128, 3, 2]], # 3-P2/4 160*160*128
# ELAN1
[-1, 1, Conv, [64, 1, 1]],
[-2, 1, Conv, [64, 1, 1]],
[-1, 1, Conv, [64, 3, 1]],
[-1, 1, Conv, [64, 3, 1]],
[-1, 1, Conv, [64, 3, 1]],
[-1, 1, Conv, [64, 3, 1]],
[[-1, -3, -5, -6], 1, Concat, [1]],
[-1, 1, Conv, [256, 1, 1]], # 11 160*160*256
# MPConv
[-1, 1, MP, []],
[-1, 1, Conv, [128, 1, 1]],
[-3, 1, Conv, [128, 1, 1]],
[-1, 1, Conv, [128, 3, 2]],
[[-1, -3], 1, Concat, [1]], # 16-P3/8 80*80*256
# ELAN1
[-1, 1, Conv, [128, 1, 1]],
[-2, 1, Conv, [128, 1, 1]],
[-1, 1, Conv, [128, 3, 1]],
[-1, 1, Conv, [128, 3, 1]],
[-1, 1, Conv, [128, 3, 1]],
[-1, 1, Conv, [128, 3, 1]],
[[-1, -3, -5, -6], 1, Concat, [1]],
[-1, 1, Conv, [512, 1, 1]], # 24 80*80*512
# MPConv
[-1, 1, MP, []],
[-1, 1, Conv, [256, 1, 1]],
[-3, 1, Conv, [256, 1, 1]],
[-1, 1, Conv, [256, 3, 2]],
[[-1, -3], 1, Concat, [1]], # 29-P4/16 40*40*512
# ELAN1
[-1, 1, Conv, [256, 1, 1]],
[-2, 1, Conv, [256, 1, 1]],
[-1, 1, Conv, [256, 3, 1]],
[-1, 1, Conv, [256, 3, 1]],
[-1, 1, Conv, [256, 3, 1]],
[-1, 1, Conv, [256, 3, 1]],
[[-1, -3, -5, -6], 1, Concat, [1]],
[-1, 1, Conv, [1024, 1, 1]], # 37 40*40*1024
# MPConv
[-1, 1, MP, []],
[-1, 1, Conv, [512, 1, 1]],
[-3, 1, Conv, [512, 1, 1]],
[-1, 1, Conv, [512, 3, 2]],
[[-1, -3], 1, Concat, [1]], # 42-P5/32 20*20*1024
# ELAN1
[-1, 1, Conv, [256, 1, 1]],
[-2, 1, Conv, [256, 1, 1]],
[-1, 1, Conv, [256, 3, 1]],
[-1, 1, Conv, [256, 3, 1]],
[-1, 1, Conv, [256, 3, 1]],
[-1, 1, Conv, [256, 3, 1]],
[[-1, -3, -5, -6], 1, Concat, [1]],
[-1, 1, Conv, [1024, 1, 1]], # 50 20*20*1024
]
# yolov7 head
head:
[[-1, 1, SPPCSPC, [512]], # 51 20*20*512
[-1, 1, Conv, [256, 1, 1]],# 20*20*256
[-1, 1, nn.Upsample, [None, 2, 'nearest']],
[37, 1, Conv, [256, 1, 1]], # route backbone P4 40*40*1024->40*40*256
[[-1, -2], 1, Concat, [1]], # 40*40*512
# ELAN2 注意：Head和Backbone的ELAN不一样
[-1, 1, Conv, [256, 1, 1]],
[-2, 1, Conv, [256, 1, 1]],
[-1, 1, Conv, [128, 3, 1]],
[-1, 1, Conv, [128, 3, 1]],
[-1, 1, Conv, [128, 3, 1]],
[-1, 1, Conv, [128, 3, 1]],
[[-1, -2, -3, -4, -5, -6], 1, Concat, [1]],
[-1, 1, Conv, [256, 1, 1]], # 63 40*40*256
[-1, 1, Conv, [128, 1, 1]], # 80*80*128
[-1, 1, nn.Upsample, [None, 2, 'nearest']],
[24, 1, Conv, [128, 1, 1]], # route backbone P3 80*80*512->80*80*128
[[-1, -2], 1, Concat, [1]],# 80*80*256
# ELAN2
[-1, 1, Conv, [128, 1, 1]],
[-2, 1, Conv, [128, 1, 1]],
[-1, 1, Conv, [64, 3, 1]],
[-1, 1, Conv, [64, 3, 1]],
[-1, 1, Conv, [64, 3, 1]],
[-1, 1, Conv, [64, 3, 1]],
[[-1, -2, -3, -4, -5, -6], 1, Concat, [1]],
[-1, 1, Conv, [128, 1, 1]], # 75 80*80*128
# MPConv Channel × 2
[-1, 1, MP, []],
[-1, 1, Conv, [128, 1, 1]],
[-3, 1, Conv, [128, 1, 1]],
[-1, 1, Conv, [128, 3, 2]],
[[-1, -3, 63], 1, Concat, [1]],# 40*40*256
# ELAN2
[-1, 1, Conv, [256, 1, 1]],
[-2, 1, Conv, [256, 1, 1]],
[-1, 1, Conv, [128, 3, 1]],
[-1, 1, Conv, [128, 3, 1]],
[-1, 1, Conv, [128, 3, 1]],
[-1, 1, Conv, [128, 3, 1]],
[[-1, -2, -3, -4, -5, -6], 1, Concat, [1]],
[-1, 1, Conv, [256, 1, 1]], # 88 40*40*256
# MPConv Channel × 2
[-1, 1, MP, []],
[-1, 1, Conv, [256, 1, 1]],
[-3, 1, Conv, [256, 1, 1]],
[-1, 1, Conv, [256, 3, 2]],
[[-1, -3, 51], 1, Concat, [1]],# 40*40*512
# ELAN2
[-1, 1, Conv, [512, 1, 1]],
[-2, 1, Conv, [512, 1, 1]],
[-1, 1, Conv, [256, 3, 1]],
[-1, 1, Conv, [256, 3, 1]],
[-1, 1, Conv, [256, 3, 1]],
[-1, 1, Conv, [256, 3, 1]],
[[-1, -2, -3, -4, -5, -6], 1, Concat, [1]],
[-1, 1, Conv, [512, 1, 1]], # 101 20*20*512
[75, 1, RepConv, [256, 3, 1]],#102 80*80*256
[88, 1, RepConv, [512, 3, 1]],#103 40*40*512
[101, 1, RepConv, [1024, 3, 1]],#104 20*20*1024
[[102,103,104], 1, IDetect, [nc, anchors]], # Detect(P3, P4, P5)
]
组件结构
CBS 模块
yaml 文件中的Conv表示卷积归一化激活
对于CBS模块，我们可以看从图中可以看出它是由一个Conv层，也就是卷积层，一个BN层，也就是Batch normalization层，还有一个Silu层，这是一个激活函数。silu激活函数是swish激活函数的变体。
python">class Conv(nn.Module):
# Standard convolution
def __init__(self, c1, c2, k=1, s=1, p=None, g=1, act=True): # ch_in, ch_out, kernel, stride, padding, groups
super(Conv, self).__init__()
self.conv = nn.Conv2d(c1, c2, k, s, autopad(k, p), groups=g, bias=False)
self.bn = nn.BatchNorm2d(c2)
self.act = nn.SiLU() if act is True else (act if isinstance(act, nn.Module) else nn.Identity())
def forward(self, x):
return self.act(self.bn(self.conv(x)))
def fuseforward(self, x):
return self.act(self.conv(x))
ELAN1
利用Conv构件围城的模块，在backbone中通道数扩张两倍
python"> #[-1, 1, Conv, [128, 3, 2]], # 3-P2/4 160*160*128
# ELAN1
[-1, 1, Conv, [64, 1, 1]],
[-2, 1, Conv, [64, 1, 1]],
[-1, 1, Conv, [64, 3, 1]],
[-1, 1, Conv, [64, 3, 1]],
[-1, 1, Conv, [64, 3, 1]],
[-1, 1, Conv, [64, 3, 1]],
[[-1, -3, -5, -6], 1, Concat, [1]],
[-1, 1, Conv, [256, 1, 1]], # 11 160*160*256
ELAN2
利用Conv构件围城的模块，在head中通道数减半
python"> # [[-1, -2], 1, Concat, [1]], #55 40*40*512
# ELAN2 注意：Head和Backbone的ELAN不一样
[-1, 1, Conv, [256, 1, 1]],
[-2, 1, Conv, [256, 1, 1]],
[-1, 1, Conv, [128, 3, 1]],
[-1, 1, Conv, [128, 3, 1]],
[-1, 1, Conv, [128, 3, 1]],
[-1, 1, Conv, [128, 3, 1]],
[[-1, -2, -3, -4, -5, -6], 1, Concat, [1]],
[-1, 1, Conv, [256, 1, 1]], # 63 40*40*256
MP1&2
MP1
[-1, 1, Conv, [128, 3, 2]], 表述输出128，
python"> #[-1, 1, Conv, [256, 1, 1]], # 11 160*160*256
# MPConv-1 backbone 下采样 通道数不变
[-1, 1, MP, []],
[-1, 1, Conv, [128, 1, 1]],
[-3, 1, Conv, [128, 1, 1]],
[-1, 1, Conv, [128, 3, 2]],
[[-1, -3], 1, Concat, [1]], # 16-P3/8 80*80*256
MP2
head部分，尺寸减半，通道数扩张为两倍
python"> # [-1, 1, Conv, [256, 1, 1]], # 88 40*40*256
# MPConv Channel × 2
[-1, 1, MP, []],
[-1, 1, Conv, [256, 1, 1]],
[-3, 1, Conv, [256, 1, 1]],
[-1, 1, Conv, [256, 3, 2]],
[[-1, -3, 51], 1, Concat, [1]],# 40*40*512
SPPCSPC
类似于yolov5中的SPPF，不同的是，使用了5×5、9×9、13×13最大池化。
python">class SPPCSPC(nn.Module):
# CSP https://github.com/WongKinYiu/CrossStagePartialNetworks
def __init__(self, c1, c2, n=1, shortcut=False, g=1, e=0.5, k=(5, 9, 13)):
super(SPPCSPC, self).__init__()
c_ = int(2 * c2 * e) # hidden channels
self.cv1 = Conv(c1, c_, 1, 1)
self.cv2 = Conv(c1, c_, 1, 1)
self.cv3 = Conv(c_, c_, 3, 1)
self.cv4 = Conv(c_, c_, 1, 1)
self.m = nn.ModuleList([nn.MaxPool2d(kernel_size=x, stride=1, padding=x // 2) for x in k])
self.cv5 = Conv(4 * c_, c_, 1, 1)
self.cv6 = Conv(c_, c_, 3, 1)
self.cv7 = Conv(2 * c_, c2, 1, 1)
def forward(self, x):
x1 = self.cv4(self.cv3(self.cv1(x)))
y1 = self.cv6(self.cv5(torch.cat([x1] + [m(x1) for m in self.m], 1)))
y2 = self.cv2(x)
return self.cv7(torch.cat((y1, y2), dim=1))
参考
【YOLOv7_0.1】网络结构与源码解析
https://blog.csdn.net/weixin_43799388/article/details/126164288
YOLOV7详细解读（一）网络架构解读
https://blog.csdn.net/qq128252/article/details/126673493
睿智的目标检测61——Pytorch搭建YoloV7目标检测平台
https://blog.csdn.net/weixin_44791964/article/details/125827160
