https://github.com/archsyscall/improvedresnet-tensorflow2
Implementation 'Improved Residual Networks for Image and Video Recognition'
https://github.com/archsyscall/improvedresnet-tensorflow2
Last synced: 4 months ago
JSON representation
Implementation 'Improved Residual Networks for Image and Video Recognition'
- Host: GitHub
- URL: https://github.com/archsyscall/improvedresnet-tensorflow2
- Owner: archsyscall
- Created: 2020-04-24T15:22:04.000Z (about 5 years ago)
- Default Branch: master
- Last Pushed: 2020-04-24T15:22:33.000Z (about 5 years ago)
- Last Synced: 2025-02-02T07:27:41.851Z (5 months ago)
- Language: Python
- Size: 2.93 KB
- Stars: 3
- Watchers: 4
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Improved Residual Networks
This is a repository that implements ["Improved Residual Networks for Image and Video Recognition"](https://arxiv.org/pdf/2004.04989.pdf) using TensorFlow2. The author's official repository is [pytorch-iresnet](https://github.com/iduta/iresnet). Most of them were referred to on [pytorch-iresnet](https://github.com/iduta/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**% |### Requirements
* tensorflow > 2.1
### Usage
```bash
$ python src/train \
--nets={iresnet18, iresnet50, etc...}
--batch_size={default=64}
--epochs={default=100}
```