| Architecture | 1x1 Convolution, Squeeze-and-Excitation Block, Batch Normalization, Convolution, Grouped Convolution, Global Average Pooling, ResNeXt Block, Residual Connection, ReLU, Max Pooling, Softmax |
|---|---|
| ID | gluon_seresnext101_32x4d |
| SHOW MORE |
| Architecture | 1x1 Convolution, Squeeze-and-Excitation Block, Batch Normalization, Convolution, Grouped Convolution, Global Average Pooling, ResNeXt Block, Residual Connection, ReLU, Max Pooling, Softmax |
|---|---|
| ID | gluon_seresnext101_64x4d |
| SHOW MORE |
| Architecture | 1x1 Convolution, Squeeze-and-Excitation Block, Batch Normalization, Convolution, Grouped Convolution, Global Average Pooling, ResNeXt Block, Residual Connection, ReLU, Max Pooling, Softmax |
|---|---|
| ID | gluon_seresnext50_32x4d |
| SHOW MORE |
SE ResNeXt is a variant of a ResNext that employs squeeze-and-excitation blocks to enable the network to perform dynamic channel-wise feature recalibration.
The weights from this model were ported from Gluon.
To load a pretrained model:
import timm
m = timm.create_model('gluon_seresnext50_32x4d', pretrained=True)
m.eval()
Replace the model name with the variant you want to use, e.g. gluon_seresnext50_32x4d. 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{hu2019squeezeandexcitation,
title={Squeeze-and-Excitation Networks},
author={Jie Hu and Li Shen and Samuel Albanie and Gang Sun and Enhua Wu},
year={2019},
eprint={1709.01507},
archivePrefix={arXiv},
primaryClass={cs.CV}
}
Image Classification
| BENCHMARK | MODEL | METRIC NAME | METRIC VALUE | GLOBAL RANK |
|---|---|---|---|---|
| ImageNet | gluon_seresnext101_64x4d | Top 1 Accuracy | 80.88% | # 72 |
| Top 5 Accuracy | 95.31% | # 72 | ||
| ImageNet | gluon_seresnext101_32x4d | Top 1 Accuracy | 80.87% | # 73 |
| Top 5 Accuracy | 95.29% | # 73 | ||
| ImageNet | gluon_seresnext50_32x4d | Top 1 Accuracy | 79.92% | # 100 |
| Top 5 Accuracy | 94.82% | # 100 |
