计算模型的一些参数
!pip install torchstat
首先安装torchstat
from torchstat import stat
# 导入所有的网络架构
# 这里是我创建的一些模型
from model.wideresnet import WideResNet
from model.wideresnet_CDD import WideResNet_CDD
from model.resnet import Resnet
from model.resnet_CDD import Resnet_CDD
from model.se_resnet import Se_net
from model.se_resnet_CDD import Se_net_CDD
from model.densenet3 import DenseNet
from model.densenet3_CDD import DenseNet_CDD
from model.shake_shake import Shake_shake
from model.shake_shake_CDD import Shake_shake_CDD
from model.pyramid import Pyramid
# model = WideResNet(40, 100, widen_factor=12, dropRate=0.0, nc=3, nd=1)
model = Resnet_CDD(layers=32, num_classes=100, nc=3, nd=1, width=4)
stat(model, ( 3, 32, 32))
只能计算不同尺寸的单张图片的计算复杂度等一些参数
若需要计算多张图片
直接用现有结果✖图片数量即可。
计算结果如下
module name input shape output shape params memory(MB) MAdd Flops MemRead(B) MemWrite(B) duration[%] MemR+W(B)
0 conv1 3 32 32 64 32 32 1728.0 0.25 3,473,408.0 1,769,472.0 19200.0 262144.0 1.57% 281344.0
1 bn1 64 32 32 64 32 32 128.0 0.25 262,144.0 131,072.0 262656.0 262144.0 0.88% 524800.0
2 layer1.0.conv1 64 32 32 64 32 32 36864.0 0.25 75,431,936.0 37,748,736.0 409600.0 262144.0 2.41% 671744.0
3 layer1.0.bn1 64 32 32 64 32 32 128.0 0.25 262,144.0 131,072.0 262656.0 262144.0 0.79% 524800.0
4 layer1.0.conv2 64 32 32 64 32 32 36864.0 0.25 75,431,936.0 37,748,736.0 409600.0 262144.0 1.75% 671744.0
5 layer1.0.bn2 64 32 32 64 32 32 128.0 0.25 262,144.0 131,072.0 262656.0 262144.0 0.77% 524800.0
6 layer1.0.shortcut 64 32 32 64 32 32 0.0 0.25 0.0 0.0 0.0 0.0 0.01% 0.0
7 layer1.1.conv1 64 32 32 64 32 32 36864.0 0.25 75,431,936.0 37,748,736.0 409600.0 262144.0 1.77% 671744.0
8 layer1.1.bn1 64 32 32 64 32 32 128.0 0.25 262,144.0 131,072.0 262656.0 262144.0 0.78% 524800.0
9 layer1.1.conv2 64 32 32 64 32 32 36864.0 0.25 75,431,936.0 37,748,736.0 409600.0 262144.0 1.73% 671744.0
10 layer1.1.bn2 64 32 32 64 32 32 128.0 0.25 262,144.0 131,072.0 262656.0 262144.0 0.74% 524800.0
11 layer1.1.shortcut 64 32 32 64 32 32 0.0 0.25 0.0 0.0 0.0 0.0 0.01% 0.0
12 layer1.2.conv1 64 32 32 64 32 32 36864.0 0.25 75,431,936.0 37,748,736.0 409600.0 262144.0 1.77% 671744.0
13 layer1.2.bn1 64 32 32 64 32 32 128.0 0.25 262,144.0 131,072.0 262656.0 262144.0 0.74% 524800.0
14 layer1.2.conv2 64 32 32 64 32 32 36864.0 0.25 75,431,936.0 37,748,736.0 409600.0 262144.0 1.87% 671744.0
15 layer1.2.bn2 64 32 32 64 32 32 128.0 0.25 262,144.0 131,072.0 262656.0 262144.0 0.77% 524800.0
16 layer1.2.shortcut 64 32 32 64 32 32 0.0 0.25 0.0 0.0 0.0 0.0 0.01% 0.0
17 layer1.3.conv1 64 32 32 64 32 32 36864.0 0.25 75,431,936.0 37,748,736.0 409600.0 262144.0 2.18% 671744.0
18 layer1.3.bn1 64 32 32 64 32 32 128.0 0.25 262,144.0 131,072.0 262656.0 262144.0 1.20% 524800.0
19 layer1.3.conv2 64 32 32 64 32 32 36864.0 0.25 75,431,936.0 37,748,736.0 409600.0 262144.0 1.89% 671744.0
20 layer1.3.bn2 64 32 32 64 32 32 128.0 0.25 262,144.0 131,072.0 262656.0 262144.0 0.75% 524800.0
21 layer1.3.shortcut 64 32 32 64 32 32 0.0 0.25 0.0 0.0 0.0 0.0 0.01% 0.0
22 layer1.4.conv1 64 32 32 64 32 32 36864.0 0.25 75,431,936.0 37,748,736.0 409600.0 262144.0 1.71% 671744.0
23 layer1.4.bn1 64 32 32 64 32 32 128.0 0.25 262,144.0 131,072.0 262656.0 262144.0 0.76% 524800.0
24 layer1.4.conv2 64 32 32 64 32 32 36864.0 0.25 75,431,936.0 37,748,736.0 409600.0 262144.0 1.72% 671744.0
25 layer1.4.bn2 64 32 32 64 32 32 128.0 0.25 262,144.0 131,072.0 262656.0 262144.0 0.76% 524800.0
26 layer1.4.shortcut 64 32 32 64 32 32 0.0 0.25 0.0 0.0 0.0 0.0 0.01% 0.0
27 CDD1.drop 64 32 32 64 32 32 0.0 0.25 0.0 0.0 0.0 0.0 0.71% 0.0
28 CDD1.norm 64 32 32 64 32 32 128.0 0.25 262,144.0 131,072.0 262656.0 262144.0 0.82% 524800.0
29 CDD1.CDD 64 32 32 64 32 32 36928.0 0.25 75,497,472.0 37,814,272.0 409856.0 262144.0 2.43% 672000.0
30 layer2.0.conv1 64 32 32 128 16 16 73728.0 0.12 37,715,968.0 18,874,368.0 557056.0 131072.0 1.95% 688128.0
31 layer2.0.bn1 128 16 16 128 16 16 256.0 0.12 131,072.0 65,536.0 132096.0 131072.0 0.69% 263168.0
32 layer2.0.conv2 128 16 16 128 16 16 147456.0 0.12 75,464,704.0 37,748,736.0 720896.0 131072.0 2.17% 851968.0
33 layer2.0.bn2 128 16 16 128 16 16 256.0 0.12 131,072.0 65,536.0 132096.0 131072.0 0.76% 263168.0
34 layer2.0.shortcut.0 64 32 32 128 16 16 8192.0 0.12 4,161,536.0 2,097,152.0 294912.0 131072.0 1.64% 425984.0
35 layer2.0.shortcut.1 128 16 16 128 16 16 256.0 0.12 131,072.0 65,536.0 132096.0 131072.0 0.70% 263168.0
36 layer2.1.conv1 128 16 16 128 16 16 147456.0 0.12 75,464,704.0 37,748,736.0 720896.0 131072.0 1.72% 851968.0
37 layer2.1.bn1 128 16 16 128 16 16 256.0 0.12 131,072.0 65,536.0 132096.0 131072.0 0.74% 263168.0
38 layer2.1.conv2 128 16 16 128 16 16 147456.0 0.12 75,464,704.0 37,748,736.0 720896.0 131072.0 1.65% 851968.0
39 layer2.1.bn2 128 16 16 128 16 16 256.0 0.12 131,072.0 65,536.0 132096.0 131072.0 0.71% 263168.0
40 layer2.1.shortcut 128 16 16 128 16 16 0.0 0.12 0.0 0.0 0.0 0.0 0.01% 0.0
41 layer2.2.conv1 128 16 16 128 16 16 147456.0 0.12 75,464,704.0 37,748,736.0 720896.0 131072.0 1.68% 851968.0
42 layer2.2.bn1 128 16 16 128 16 16 256.0 0.12 131,072.0 65,536.0 132096.0 131072.0 0.75% 263168.0
43 layer2.2.conv2 128 16 16 128 16 16 147456.0 0.12 75,464,704.0 37,748,736.0 720896.0 131072.0 1.77% 851968.0
44 layer2.2.bn2 128 16 16 128 16 16 256.0 0.12 131,072.0 65,536.0 132096.0 131072.0 0.70% 263168.0
45 layer2.2.shortcut 128 16 16 128 16 16 0.0 0.12 0.0 0.0 0.0 0.0 0.01% 0.0
46 layer2.3.conv1 128 16 16 128 16 16 147456.0 0.12 75,464,704.0 37,748,736.0 720896.0 131072.0 1.68% 851968.0
47 layer2.3.bn1 128 16 16 128 16 16 256.0 0.12 131,072.0 65,536.0 132096.0 131072.0 0.74% 263168.0
48 layer2.3.conv2 128 16 16 128 16 16 147456.0 0.12 75,464,704.0 37,748,736.0 720896.0 131072.0 1.66% 851968.0
49 layer2.3.bn2 128 16 16 128 16 16 256.0 0.12 131,072.0 65,536.0 132096.0 131072.0 0.72% 263168.0
50 layer2.3.shortcut 128 16 16 128 16 16 0.0 0.12 0.0 0.0 0.0 0.0 0.01% 0.0
51 layer2.4.conv1 128 16 16 128 16 16 147456.0 0.12 75,464,704.0 37,748,736.0 720896.0 131072.0 1.70% 851968.0
52 layer2.4.bn1 128 16 16 128 16 16 256.0 0.12 131,072.0 65,536.0 132096.0 131072.0 0.71% 263168.0
53 layer2.4.conv2 128 16 16 128 16 16 147456.0 0.12 75,464,704.0 37,748,736.0 720896.0 131072.0 1.66% 851968.0
54 layer2.4.bn2 128 16 16 128 16 16 256.0 0.12 131,072.0 65,536.0 132096.0 131072.0 0.71% 263168.0
55 layer2.4.shortcut 128 16 16 128 16 16 0.0 0.12 0.0 0.0 0.0 0.0 0.01% 0.0
56 CDD2.drop 128 16 16 128 16 16 0.0 0.12 0.0 0.0 0.0 0.0 0.70% 0.0
57 CDD2.norm 128 16 16 128 16 16 256.0 0.12 131,072.0 65,536.0 132096.0 131072.0 0.76% 263168.0
58 CDD2.CDD 64 32 32 128 16 16 73856.0 0.12 37,748,736.0 18,907,136.0 557568.0 131072.0 1.92% 688640.0
59 layer3.0.conv1 128 16 16 256 8 8 294912.0 0.06 37,732,352.0 18,874,368.0 1310720.0 65536.0 2.11% 1376256.0
60 layer3.0.bn1 256 8 8 256 8 8 512.0 0.06 65,536.0 32,768.0 67584.0 65536.0 0.69% 133120.0
61 layer3.0.conv2 256 8 8 256 8 8 589824.0 0.06 75,481,088.0 37,748,736.0 2424832.0 65536.0 2.85% 2490368.0
62 layer3.0.bn2 256 8 8 256 8 8 512.0 0.06 65,536.0 32,768.0 67584.0 65536.0 0.67% 133120.0
63 layer3.0.shortcut.0 128 16 16 256 8 8 32768.0 0.06 4,177,920.0 2,097,152.0 262144.0 65536.0 1.62% 327680.0
64 layer3.0.shortcut.1 256 8 8 256 8 8 512.0 0.06 65,536.0 32,768.0 67584.0 65536.0 0.67% 133120.0
65 layer3.1.conv1 256 8 8 256 8 8 589824.0 0.06 75,481,088.0 37,748,736.0 2424832.0 65536.0 2.16% 2490368.0
66 layer3.1.bn1 256 8 8 256 8 8 512.0 0.06 65,536.0 32,768.0 67584.0 65536.0 0.69% 133120.0
67 layer3.1.conv2 256 8 8 256 8 8 589824.0 0.06 75,481,088.0 37,748,736.0 2424832.0 65536.0 2.05% 2490368.0
68 layer3.1.bn2 256 8 8 256 8 8 512.0 0.06 65,536.0 32,768.0 67584.0 65536.0 0.70% 133120.0
69 layer3.1.shortcut 256 8 8 256 8 8 0.0 0.06 0.0 0.0 0.0 0.0 0.01% 0.0
70 layer3.2.conv1 256 8 8 256 8 8 589824.0 0.06 75,481,088.0 37,748,736.0 2424832.0 65536.0 2.07% 2490368.0
71 layer3.2.bn1 256 8 8 256 8 8 512.0 0.06 65,536.0 32,768.0 67584.0 65536.0 0.69% 133120.0
72 layer3.2.conv2 256 8 8 256 8 8 589824.0 0.06 75,481,088.0 37,748,736.0 2424832.0 65536.0 2.05% 2490368.0
73 layer3.2.bn2 256 8 8 256 8 8 512.0 0.06 65,536.0 32,768.0 67584.0 65536.0 0.71% 133120.0
74 layer3.2.shortcut 256 8 8 256 8 8 0.0 0.06 0.0 0.0 0.0 0.0 0.01% 0.0
75 layer3.3.conv1 256 8 8 256 8 8 589824.0 0.06 75,481,088.0 37,748,736.0 2424832.0 65536.0 2.10% 2490368.0
76 layer3.3.bn1 256 8 8 256 8 8 512.0 0.06 65,536.0 32,768.0 67584.0 65536.0 0.70% 133120.0
77 layer3.3.conv2 256 8 8 256 8 8 589824.0 0.06 75,481,088.0 37,748,736.0 2424832.0 65536.0 1.99% 2490368.0
78 layer3.3.bn2 256 8 8 256 8 8 512.0 0.06 65,536.0 32,768.0 67584.0 65536.0 0.68% 133120.0
79 layer3.3.shortcut 256 8 8 256 8 8 0.0 0.06 0.0 0.0 0.0 0.0 0.01% 0.0
80 layer3.4.conv1 256 8 8 256 8 8 589824.0 0.06 75,481,088.0 37,748,736.0 2424832.0 65536.0 2.05% 2490368.0
81 layer3.4.bn1 256 8 8 256 8 8 512.0 0.06 65,536.0 32,768.0 67584.0 65536.0 0.66% 133120.0
82 layer3.4.conv2 256 8 8 256 8 8 589824.0 0.06 75,481,088.0 37,748,736.0 2424832.0 65536.0 2.02% 2490368.0
83 layer3.4.bn2 256 8 8 256 8 8 512.0 0.06 65,536.0 32,768.0 67584.0 65536.0 0.73% 133120.0
84 layer3.4.shortcut 256 8 8 256 8 8 0.0 0.06 0.0 0.0 0.0 0.0 0.01% 0.0
85 CDD3.drop 256 8 8 256 8 8 0.0 0.06 0.0 0.0 0.0 0.0 0.65% 0.0
86 CDD3.norm 256 8 8 256 8 8 512.0 0.06 65,536.0 32,768.0 67584.0 65536.0 1.93% 133120.0
87 CDD3.CDD 128 16 16 256 8 8 295168.0 0.06 37,748,736.0 18,890,752.0 1311744.0 65536.0 2.11% 1377280.0
88 linear 256 100 25700.0 0.00 51,100.0 25,600.0 103824.0 400.0 0.67% 104224.0
total 7857892.0 12.94 2,356,643,740.0 1,179,067,392.0 103824.0 400.0 100.00% 53793056.0
=============================================================================================================================================================
Total params: 7,857,892
-------------------------------------------------------------------------------------------------------------------------------------------------------------
Total memory: 12.94MB
Total MAdd: 2.36GMAdd
Total Flops: 1.18GFlops
Total MemR+W: 51.3MB
扫一扫
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
