Architecture | 1x1 Convolution, Bottleneck Residual Block, Batch Normalization, Convolution, Global Average Pooling, Residual Block, Residual Connection, ReLU, Max Pooling, Softmax |
---|---|
ID | resnet101d |
SHOW MORE |
Architecture | 1x1 Convolution, Bottleneck Residual Block, Batch Normalization, Convolution, Global Average Pooling, Residual Block, Residual Connection, ReLU, Max Pooling, Softmax |
---|---|
ID | resnet152d |
SHOW MORE |
Architecture | 1x1 Convolution, Bottleneck Residual Block, Batch Normalization, Convolution, Global Average Pooling, Residual Block, Residual Connection, ReLU, Max Pooling, Softmax |
---|---|
ID | resnet18d |
SHOW MORE |
Architecture | 1x1 Convolution, Bottleneck Residual Block, Batch Normalization, Convolution, Global Average Pooling, Residual Block, Residual Connection, ReLU, Max Pooling, Softmax |
---|---|
ID | resnet200d |
SHOW MORE |
Architecture | 1x1 Convolution, Bottleneck Residual Block, Batch Normalization, Convolution, Global Average Pooling, Residual Block, Residual Connection, ReLU, Max Pooling, Softmax |
---|---|
ID | resnet26d |
SHOW MORE |
Architecture | 1x1 Convolution, Bottleneck Residual Block, Batch Normalization, Convolution, Global Average Pooling, Residual Block, Residual Connection, ReLU, Max Pooling, Softmax |
---|---|
ID | resnet34d |
SHOW MORE |
Architecture | 1x1 Convolution, Bottleneck Residual Block, Batch Normalization, Convolution, Global Average Pooling, Residual Block, Residual Connection, ReLU, Max Pooling, Softmax |
---|---|
ID | resnet50d |
SHOW MORE |
ResNet-D is a modification on the ResNet architecture that utilises an average pooling tweak for downsampling. The motivation is that in the unmodified ResNet, the 1×1 convolution for the downsampling block ignores 3/4 of input feature maps, so this is modified so no information will be ignored
To load a pretrained model:
import timm
m = timm.create_model('resnet50d', pretrained=True)
m.eval()
Replace the model name with the variant you want to use, e.g. resnet50d
. You can find the IDs in the model summaries at the top of this page.
You can follow the timm recipe scripts for training a new model afresh.
@misc{he2018bag,
title={Bag of Tricks for Image Classification with Convolutional Neural Networks},
author={Tong He and Zhi Zhang and Hang Zhang and Zhongyue Zhang and Junyuan Xie and Mu Li},
year={2018},
eprint={1812.01187},
archivePrefix={arXiv},
primaryClass={cs.CV}
}
MODEL | TOP 1 ACCURACY | TOP 5 ACCURACY |
---|---|---|
resnet200d | 83.24% | 96.49% |
resnet152d | 83.13% | 96.35% |
resnet101d | 82.31% | 96.06% |
resnet50d | 80.55% | 95.16% |
resnet34d | 77.11% | 93.38% |
resnet26d | 76.69% | 93.15% |
resnet18d | 72.27% | 90.69% |