Gloun ResNet

Last updated on Feb 14, 2021

gluon_resnet101_v1b

Parameters 45 Million
FLOPs 10 Billion
File Size 170.44 MB
Training Data ImageNet
Training Resources
Training Time

Architecture 1x1 Convolution, Bottleneck Residual Block, Batch Normalization, Convolution, Global Average Pooling, Residual Block, Residual Connection, ReLU, Max Pooling, Softmax
ID gluon_resnet101_v1b
Crop Pct 0.875
Image Size 224
Interpolation bicubic
SHOW MORE
SHOW LESS
gluon_resnet101_v1c

Parameters 45 Million
FLOPs 10 Billion
File Size 170.52 MB
Training Data ImageNet
Training Resources
Training Time

Architecture 1x1 Convolution, Bottleneck Residual Block, Batch Normalization, Convolution, Global Average Pooling, Residual Block, Residual Connection, ReLU, Max Pooling, Softmax
ID gluon_resnet101_v1c
Crop Pct 0.875
Image Size 224
Interpolation bicubic
SHOW MORE
SHOW LESS
gluon_resnet101_v1d

Parameters 45 Million
FLOPs 10 Billion
File Size 170.52 MB
Training Data ImageNet
Training Resources
Training Time

Architecture 1x1 Convolution, Bottleneck Residual Block, Batch Normalization, Convolution, Global Average Pooling, Residual Block, Residual Connection, ReLU, Max Pooling, Softmax
ID gluon_resnet101_v1d
Crop Pct 0.875
Image Size 224
Interpolation bicubic
SHOW MORE
SHOW LESS
gluon_resnet101_v1s

Parameters 45 Million
FLOPs 12 Billion
File Size 170.92 MB
Training Data ImageNet
Training Resources
Training Time

Architecture 1x1 Convolution, Bottleneck Residual Block, Batch Normalization, Convolution, Global Average Pooling, Residual Block, Residual Connection, ReLU, Max Pooling, Softmax
ID gluon_resnet101_v1s
Crop Pct 0.875
Image Size 224
Interpolation bicubic
SHOW MORE
SHOW LESS
gluon_resnet152_v1b

Parameters 60 Million
FLOPs 15 Billion
File Size 230.34 MB
Training Data ImageNet
Training Resources
Training Time

Architecture 1x1 Convolution, Bottleneck Residual Block, Batch Normalization, Convolution, Global Average Pooling, Residual Block, Residual Connection, ReLU, Max Pooling, Softmax
ID gluon_resnet152_v1b
Crop Pct 0.875
Image Size 224
Interpolation bicubic
SHOW MORE
SHOW LESS
gluon_resnet152_v1c

Parameters 60 Million
FLOPs 15 Billion
File Size 230.42 MB
Training Data ImageNet
Training Resources
Training Time

Architecture 1x1 Convolution, Bottleneck Residual Block, Batch Normalization, Convolution, Global Average Pooling, Residual Block, Residual Connection, ReLU, Max Pooling, Softmax
ID gluon_resnet152_v1c
Crop Pct 0.875
Image Size 224
Interpolation bicubic
SHOW MORE
SHOW LESS
gluon_resnet152_v1d

Parameters 60 Million
FLOPs 15 Billion
File Size 230.42 MB
Training Data ImageNet
Training Resources
Training Time

Architecture 1x1 Convolution, Bottleneck Residual Block, Batch Normalization, Convolution, Global Average Pooling, Residual Block, Residual Connection, ReLU, Max Pooling, Softmax
ID gluon_resnet152_v1d
Crop Pct 0.875
Image Size 224
Interpolation bicubic
SHOW MORE
SHOW LESS
gluon_resnet152_v1s

Parameters 60 Million
FLOPs 17 Billion
File Size 230.82 MB
Training Data ImageNet
Training Resources
Training Time

Architecture 1x1 Convolution, Bottleneck Residual Block, Batch Normalization, Convolution, Global Average Pooling, Residual Block, Residual Connection, ReLU, Max Pooling, Softmax
ID gluon_resnet152_v1s
Crop Pct 0.875
Image Size 224
Interpolation bicubic
SHOW MORE
SHOW LESS
gluon_resnet18_v1b

Parameters 12 Million
FLOPs 2 Billion
File Size 44.65 MB
Training Data ImageNet
Training Resources
Training Time

Architecture 1x1 Convolution, Bottleneck Residual Block, Batch Normalization, Convolution, Global Average Pooling, Residual Block, Residual Connection, ReLU, Max Pooling, Softmax
ID gluon_resnet18_v1b
Crop Pct 0.875
Image Size 224
Interpolation bicubic
SHOW MORE
SHOW LESS
gluon_resnet34_v1b

Parameters 22 Million
FLOPs 5 Billion
File Size 83.25 MB
Training Data ImageNet
Training Resources
Training Time

Architecture 1x1 Convolution, Bottleneck Residual Block, Batch Normalization, Convolution, Global Average Pooling, Residual Block, Residual Connection, ReLU, Max Pooling, Softmax
ID gluon_resnet34_v1b
Crop Pct 0.875
Image Size 224
Interpolation bicubic
SHOW MORE
SHOW LESS
gluon_resnet50_v1b

Parameters 26 Million
FLOPs 5 Billion
File Size 97.75 MB
Training Data ImageNet
Training Resources
Training Time

Architecture 1x1 Convolution, Bottleneck Residual Block, Batch Normalization, Convolution, Global Average Pooling, Residual Block, Residual Connection, ReLU, Max Pooling, Softmax
ID gluon_resnet50_v1b
Crop Pct 0.875
Image Size 224
Interpolation bicubic
SHOW MORE
SHOW LESS
gluon_resnet50_v1c

Parameters 26 Million
FLOPs 6 Billion
File Size 97.82 MB
Training Data ImageNet
Training Resources
Training Time

Architecture 1x1 Convolution, Bottleneck Residual Block, Batch Normalization, Convolution, Global Average Pooling, Residual Block, Residual Connection, ReLU, Max Pooling, Softmax
ID gluon_resnet50_v1c
Crop Pct 0.875
Image Size 224
Interpolation bicubic
SHOW MORE
SHOW LESS
gluon_resnet50_v1d

Parameters 26 Million
FLOPs 6 Billion
File Size 97.82 MB
Training Data ImageNet
Training Resources
Training Time

Architecture 1x1 Convolution, Bottleneck Residual Block, Batch Normalization, Convolution, Global Average Pooling, Residual Block, Residual Connection, ReLU, Max Pooling, Softmax
ID gluon_resnet50_v1d
Crop Pct 0.875
Image Size 224
Interpolation bicubic
SHOW MORE
SHOW LESS
gluon_resnet50_v1s

Parameters 26 Million
FLOPs 7 Billion
File Size 98.22 MB
Training Data ImageNet
Training Resources
Training Time

Architecture 1x1 Convolution, Bottleneck Residual Block, Batch Normalization, Convolution, Global Average Pooling, Residual Block, Residual Connection, ReLU, Max Pooling, Softmax
ID gluon_resnet50_v1s
Crop Pct 0.875
Image Size 224
Interpolation bicubic
SHOW MORE
SHOW LESS
README.md

Summary

Residual Networks, or ResNets, learn residual functions with reference to the layer inputs, instead of learning unreferenced functions. Instead of hoping each few stacked layers directly fit a desired underlying mapping, residual nets let these layers fit a residual mapping. They stack residual blocks ontop of each other to form network: e.g. a ResNet-50 has fifty layers using these blocks.

The weights from this model were ported from Gluon.

How do I load this model?

To load a pretrained model:

import timm
m = timm.create_model('gluon_resnet18_v1b', pretrained=True)
m.eval()

Replace the model name with the variant you want to use, e.g. gluon_resnet18_v1b. You can find the IDs in the model summaries at the top of this page.

How do I train this model?

You can follow the timm recipe scripts for training a new model afresh.

Citation

@article{DBLP:journals/corr/HeZRS15,
  author    = {Kaiming He and
               Xiangyu Zhang and
               Shaoqing Ren and
               Jian Sun},
  title     = {Deep Residual Learning for Image Recognition},
  journal   = {CoRR},
  volume    = {abs/1512.03385},
  year      = {2015},
  url       = {http://arxiv.org/abs/1512.03385},
  archivePrefix = {arXiv},
  eprint    = {1512.03385},
  timestamp = {Wed, 17 Apr 2019 17:23:45 +0200},
  biburl    = {https://dblp.org/rec/journals/corr/HeZRS15.bib},
  bibsource = {dblp computer science bibliography, https://dblp.org}
}

Results

Image Classification on ImageNet

Image Classification on ImageNet
MODEL TOP 1 ACCURACY TOP 5 ACCURACY
gluon_resnet152_v1s 81.02% 95.42%
gluon_resnet152_v1d 80.48% 95.2%
gluon_resnet101_v1d 80.4% 95.02%
gluon_resnet101_v1s 80.29% 95.16%
gluon_resnet152_v1c 79.91% 94.85%
gluon_resnet152_v1b 79.69% 94.73%
gluon_resnet101_v1c 79.53% 94.59%
gluon_resnet101_v1b 79.3% 94.53%
gluon_resnet50_v1d 79.06% 94.46%
gluon_resnet50_v1s 78.7% 94.25%
gluon_resnet50_v1c 78.01% 93.99%
gluon_resnet50_v1b 77.58% 93.72%
gluon_resnet34_v1b 74.59% 92.0%
gluon_resnet18_v1b 70.84% 89.76%