Versions Compared

Old Version 21

changes.mady.by.user Patric Zhao

Saved on

compared with

New Version 22

changes.mady.by.user Patric Zhao

Saved on

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

Note: The dataset used ImageNet1k valdata/ are generated by https://github.com/apache/incubator-mxnet/blob/master/example/image-classification/data/imagenet1k-val.sh

Inference Accuracy Comparison
AliasNetwork
# Parameters
CPU (without MKL-DNN)CPU (with MKL-DNN) Backend
)
Delta
 top1 top5 top1 top5top1top5
alexnetAlexNet
61,100,840
0.563125000.789921880.
563125
563125000.789921880.
563125
000000000.
7899219
00000000
densenet121DenseNet-121
8,062,504
0.742031250.919296880.742031250.919296880.
7420313
000000000.
9192969
00000000
densenet161DenseNet-161
28,900,936
0.771953130.933906250.771953130.933906250.
7719531
000000000.
9339063
00000000
densenet169DenseNet-169
14,307,880
0.757109380.928281250.757109380.928281250.
7571094
000000000.
9282813
00000000
densenet201DenseNet-201
20,242,984
0.769062500.930937500.
7690625
769062500.
9309375
930937500.
7690625
000000000.
9309375
00000000
inceptionv3Inception V3 299x299
23,869,000
0.776093750.936640630.776093750.936640630.
7760938
000000000.
9366406
00000000
mobilenet0.25MobileNet 0.25
475,544
0.510390630.756875000.510390630.
756875
756875000.
5103906
000000000.
756875
00000000
mobilenet0.5MobileNet 0.5
1,342,536
0.618515630.837890630.618515630.837890630.
6185156
000000000.
8378906
00000000
mobilenet0.75MobileNet 0.75
2,601,976
0.665468750.870703130.665468750.870703130.
6654688
000000000.
8707031
00000000
mobilenet1.0MobileNet 1.0
4,253,864
0.700937500.891093750.
7009375
700937500.891093750.
7009375
000000000.
8910938
00000000
mobilenetv2_1.0MobileNetV2 1.0
3,539,136
0.699765630.892812500.699765630.
8928125
892812500.
6997656
000000000.
8928125
00000000
mobilenetv2_0.75MobileNetV2 0.75
2,653,864
0.682109380.880078130.682109380.880078130.
6821094
000000000.
8800781
00000000
mobilenetv2_0.5MobileNetV2 0.5
1,983,104
0.644531250.849296880.644531250.849296880.
6445313
000000000.
8492969
00000000
mobilenetv2_0.25MobileNetV2 0.25
1,526,856
0.508906250.745468750.508906250.745468750.
5089063
000000000.
7454688
00000000
resnet18_v1ResNet-18 V1
11,699,112
0.708125000.894531250.
708125
708125000.894531250.
708125
000000000.
8945313
00000000
resnet34_v1ResNet-34 V1
21,814,696
0.739609380.916093750.739609380.916093750.
7396094
000000000.
9160938
00000000
resnet50_v1ResNet-50 V1
25,629,032
0.760625000.930468750.
760625
760625000.930468750.
760625
000000000.
9304688
00000000
resnet101_v1ResNet-101 V1
44,695,144
0.779375000.936171880.
779375
779375000.936171880.
779375
000000000.
9361719
00000000
resnet152_v1ResNet-152 V1
60,404,072
0.783203130.938671880.783203130.938671880.
7832031
000000000.
9386719
00000000
resnet18_v2ResNet-18 V2
11,695,796
0.710468750.896718750.710468750.896718750.
7104688
000000000.
8967188
00000000
resnet34_v2ResNet-34 V2
21,811,380
0.740859380.915781250.740859380.915781250.
7408594
000000000.
9157813
00000000
resnet50_v2ResNet-50 V2
25,595,060
0.767500000.931875000.
7675
767500000.
931875
931875000.
7675
000000000.
931875
00000000
resnet101_v2ResNet-101 V2
44,639,412
0.781250000.940156250.
78125
781250000.940156250.
78125
000000000.
9401563
00000000
resnet152_v2ResNet-152 V2
60,329,140
0.785546880.941406250.785546880.941406250.
7855469
000000000.
9414063
00000000
squeezenet1.0SqueezeNet 1.0
1,248,424
0.572734380.795546880.572734380.795546880.
5727344
000000000.
7955469
00000000
squeezenet1.1SqueezeNet 1.1
1,235,496
0.570234380.796015630.570234380.796015630.
5702344
000000000.
7960156
00000000
vgg11VGG-11
132,863,336
0.670625000.875312500.
670625
670625000.
8753125
875312500.
670625
000000000.
8753125
00000000
vgg13VGG-13
133,047,848
0.681328130.879843750.681328130.879843750.
6813281
000000000.
8798438
00000000
vgg16VGG-16
138,357,544
0.720625000.905859380.
720625
720625000.905859380.
720625
000000000.
9058594
00000000
vgg19VGG-19
143,667,240
0.734687500.910000000.
7346875
734687500.
91
910000000.
7346875
000000000.
91
00000000
vgg11_bnVGG-11 with batch normalization
132,874,344
0.689531250.888828130.689531250.888828130.
6895313
000000000.
8888281
00000000
vgg13_bnVGG-13 with batch normalization
133,059,624
0.698359380.889531250.698359380.889531250.
6983594
000000000.
8895313
00000000
vgg16_bnVGG-16 with batch normalization
138,374,440
0.722265630.903906250.722265630.903906250.
7222656
000000000.
9039063
00000000
vgg19_bnVGG-19 with batch normalization
143,689,256
0.729921880.909921880.729921880.909921880.
7299219
000000000.
9099219
00000000