Skip to content

Latest commit

 

History

History
117 lines (97 loc) · 7.74 KB

README.md

File metadata and controls

117 lines (97 loc) · 7.74 KB
Raw

　　　　　　常见模型参数量、计算量和速度统计

本文总结常见模型参数量、计算量和速度信息备查

目录

[1. 分类网络](# 分类网络)

[2.目标检测网络](# 目标检测网络)

[3.关键点检测网络](# 关键点检测网络)

[4. 参考资料](# 参考资料)

分类网络

Model Input Resolution Params(M) MACs(GFLOPS) Acc@1 Acc@5
AlexNet 224x224 61.10 0.72 56.432 79.194
GoogLeNet 224x224 7.00 1.60
VGG11 224x224 132.86 7.63 69.02 88.63
VGG11_bn 224x224 132.87 7.64 70.38 89.81
VGG13 224x224 133.05 11.34 69.93 89.25
VGG13_bn 224x224 133.05 11.36 71.55 90.37
VGG16 224x224 138.36 15.5 71.59 90.38
VGG16_bn 224x224 138.37 15.53 73.37 91.50
VGG19 224x224 143.67 19.67 72.38 90.88
VGG19_bn 224x224 143.68 19.70 74.24 91.85
ResNet18 224x224 11.69 1.82 69.76 91.08
ResNet34 224x224 21.80 3.68 73.30 91.42
ResNet50 224x224 25.56 4.12 76.15 92.87
Wide_ResNet50_2 68.88 11.46
ResNet101 224x224 44.55 7.85 77.37 93.56
Wide_ResNet101_2 126.89 22.84
ResNet152 224x224 60.19 11.58 78.31 94.06
SqueezeNet1_0 224x224 1.25 0.83 58.10 80.42
SqueezeNet1_1 224x224 1.24 0.36 58.19 80.62
DenseNet121 224x224 7.98 2.88 74.65 92.17
DenseNet161 224x224 28.68 7.82 77.65 93.80
DenseNet169 224x224 14.15 3.42 76.00 93.00
DenseNet201 224x224 20.01 4.37 77.20 93.57
Inception_v3 224x224 27.16 2.85 77.45 93.56
Inception_v3 299x299 27.16 5.73 77.294 93.454
Inception_v4 299x299 42.68 13.31 80.062 94.926
Inception_bn 224x224 11.3 2.05 73.524 91.562
DPN107 224x224 86.92 18.42 79.746 94.684
DPN131 224x224 79.25 16.13 79.432 94.574
DPN68 224x224 12.61 2.36 75.868 92.774
DPN68b 224x224 12.61 2.36 77.034 93.59
DPN92 224x224 37.67 6.56 79.4 94.62
DPN98 224x224 61.57 11.76 79.224 94.488
FBResNet152 224x224 60.27 11.6 77.386 93.594
InceptionResNetv2 299x299 55.84 13.22 80.17 95.234
NasNetaLarge 331x331 88.75 24.04 82.566 96.086
NasNetaMobile 224x224 5.29 0.59 74.08 91.74
Pnasnet5Large 331x331 86.06 25.21 82.736 95.992
PolyNet 331x331 95.37 34.90 81.002 95.624
ResNeXt50_32x4d 25.03 4.29
ResNeXt101_32x4d 224x224 44.18 8.03 78.188 93.886
ResNeXt101_32x8d 88.79 16.54
ResNeXt101_64x4d 224x224 83.46 15.55 78.956 94.252
se_ResNet101 224x224 49.33 7.63 78.396 94.258
se_ResNet152 224x224 66.82 11.37 78.658 94.374
se_ResNet50 224x224 28.09 3.9 77.636 93.752
se_ResNeXt101_32x4d 224x224 48.96 8.05 80.236 95.028
se_ResNeXt50_32x4d 224x224 27.56 4.28 79.076 94.434
SENet154 224x224 115.09 20.82 81.304 95.498
Xception 299x299 22.86 8.42 78.888 94.292
TextCNN 0.15 0.009
MnasNet0_5 2.22 0.14
MnasNet0_75 3.17 0.24
MnasNet1_0 4.38 0.34
MnasNet1_3 6.28 0.53
MobileNet_v2 3.50 0.33
Shufflenet_v2_x0_5 1.37 0.05
Shufflenet_v2_x1_0 2.28 0.15
Shufflenet_v2_x1_5 3.50 0.31
Shufflenet_v2_x2_0 7.39 0.60

目标检测网络

Model Input Resolution Params Memory Feature Memory FLOPS
RFCN-Res50-Pascal 600x850 122 MB 1 GB 79 GFLOPS
RFCN-Res101-Pascal 600x850 194 MB 2 GB 117 GFLOPS
SSD-Pascal-VGGvd-300 300x300 100 MB 116 MB 31 GFLOPS
SSD-Pascal-VGGvd-512 512x512 104 MB 337 MB 91 GFLOPS
SSD-Pascal-MobileNet-ft 300x300 22 MB 37 MB 1 GFLOPS
FasterRCNN-VGGvd-Pascal 600x850 523 MB 600 MB 172 GFLOPS

分割网络

Model Input Resolution Params Memory Feature Memory FLOPS
Pascal-FCN-32s 384x384 519 MB 423 MB 125 GFLOPS
Pascal-FCN-16s 384x384 514 MB 424 MB 125 GFLOPS
Pascal-FCN-8s 384x384 513 MB 426 MB 125 GFLOPS
Deeplab-VGGvd-v2 513x513 144 MB 755 MB 202 GFLOPS
Deeplab-Res101-v2 513x513 505 MB 4 GB 346 GFLOPS

关键点检测网络

Model Input Resolution Params Memory Feature Memory FLOPS
MultiPose-mpi 368x368 196 MB 245 MB 134 GFLOPS
MultiPose-coco 368x368 200 MB 246 MB 136 GFLOPS

参考资料

pytorch-OpCounter

flops-counter.pytorch

convnet-burden

Characterization and Benchmarking of Deep Learning