This is a repository that implements "Improved Residual Networks for Image and Video Recognition" using TensorFlow2. The author's official repository is pytorch-iresnet. Most of them were referred to on pytorch-iresnet.
The accuracy on ImageNet (Data in the official repository):
| Network | 50-layers | 101-layers | 152-layers | 200-layers |
|---|---|---|---|---|
| ResNet | 76.12% | 78.00% | 78.45% | 77.55% |
| iResnet | 77.31% | 78.64% | 79.34% | 79.48% |
- tensorflow > 2.1
$ python src/train \
--nets={iresnet18, iresnet50, etc...}
--batch_size={default=64}
--epochs={default=100}