Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/XingangPan/IBN-Net
Instance-Batch Normalization Networks (ECCV2018)
https://github.com/XingangPan/IBN-Net
Last synced: about 2 months ago
JSON representation
Instance-Batch Normalization Networks (ECCV2018)
- Host: GitHub
- URL: https://github.com/XingangPan/IBN-Net
- Owner: XingangPan
- License: mit
- Created: 2018-07-20T13:02:33.000Z (about 6 years ago)
- Default Branch: master
- Last Pushed: 2023-07-04T03:50:44.000Z (about 1 year ago)
- Last Synced: 2024-07-10T09:52:05.032Z (3 months ago)
- Language: Python
- Size: 1.18 MB
- Stars: 795
- Watchers: 20
- Forks: 114
- Open Issues: 14
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
## Instance-Batch Normalization Network
### Paper
Xingang Pan, Ping Luo, Jianping Shi, Xiaoou Tang. ["Two at Once: Enhancing Learning and Generalization Capacities via IBN-Net"](https://arxiv.org/abs/1807.09441), ECCV2018.
### Introduction
- IBN-Net is a CNN model with domain/appearance invariance. It carefully unifies instance normalization and batch normalization in a single deep network.
- It provides a simple way to increase both modeling and generalization capacity without adding model complexity.
- IBN-Net is especially suitable for cross domain or person/vehicle re-identification tasks, see [michuanhaohao/reid-strong-baseline](https://github.com/michuanhaohao/reid-strong-baseline) and [strong baseline for ReID](https://arxiv.org/pdf/1906.08332.pdf) for more details.
### Requirements
- Pytorch 0.4.1 or higher
### Results
Top1/Top5 error on the ImageNet validation set are reported. You may get different results when training your models with different random seed.
| Model | origin | re-implementation | IBN-Net |
| ------------------- | ------------------ | ------------------ | ------------------ |
| DenseNet-121 | 25.0/- | 24.96/7.85 | 24.47/7.25 [[pre-trained model]](https://github.com/XingangPan/IBN-Net/releases/download/v1.0/densenet121_ibn_a-e4af5cc1.pth) |
| DenseNet-169 | 23.6/- | 24.02/7.06 | 23.25/6.51 [[pre-trained model]](https://github.com/XingangPan/IBN-Net/releases/download/v1.0/densenet169_ibn_a-9f32c161.pth) |
| ResNet-18 | - | 30.24/10.92 | 29.17/10.24 [[pre-trained model]](https://github.com/XingangPan/IBN-Net/releases/download/v1.0/resnet18_ibn_a-2f571257.pth) |
| ResNet-34 | - | 26.70/8.58 | 25.78/8.19 [[pre-trained model]](https://github.com/XingangPan/IBN-Net/releases/download/v1.0/resnet34_ibn_a-94bc1577.pth) |
| ResNet-50 | 24.7/7.8 | 24.27/7.08 | 22.54/6.32 [[pre-trained model]](https://github.com/XingangPan/IBN-Net/releases/download/v1.0/resnet50_ibn_a-d9d0bb7b.pth) |
| ResNet-101 | 23.6/7.1 | 22.48/6.23 | 21.39/5.59 [[pre-trained model]](https://github.com/XingangPan/IBN-Net/releases/download/v1.0/resnet101_ibn_a-59ea0ac6.pth) |
| ResNeXt-101 | 21.2/5.6 | 21.31/5.74 | 20.88/5.42 [[pre-trained model]](https://github.com/XingangPan/IBN-Net/releases/download/v1.0/resnext101_ibn_a-6ace051d.pth) |
| SE-ResNet-101 | 22.38/6.07 | 21.68/5.88 | 21.25/5.51 [[pre-trained model]](https://github.com/XingangPan/IBN-Net/releases/download/v1.0/se_resnet101_ibn_a-fabed4e2.pth) |
The rank1/mAP on two Re-ID benchmarks Market1501 and DukeMTMC-reID (from [michuanhaohao/reid-strong-baseline](https://github.com/michuanhaohao/reid-strong-baseline)):
| Backbone | Market1501 | DukeMTMC-reID |
| --- | -- | -- |
| ResNet50 | 94.5 (85.9) | 86.4 (76.4) |
| ResNet101 | 94.5 (87.1) | 87.6 (77.6) |
| SeResNet50 | 94.4 (86.3) | 86.4 (76.5) |
| SeResNet101 | 94.6 (87.3) | 87.5 (78.0) |
| SeResNeXt50 | 94.9 (87.6) | 88.0 (78.3) |
| SeResNeXt101 | 95.0 (88.0) | 88.4 (79.0) |
| IBN-Net-a | 95.0 (88.2) | 90.1 (79.1) |
### Load IBN-Net from torch.hub
```python
import torch
model = torch.hub.load('XingangPan/IBN-Net', 'resnet50_ibn_a', pretrained=True)
```
### Testing/Training on ImageNet
1. Clone the repository
```Shell
git clone https://github.com/XingangPan/IBN-Net.git
```
2. Download [ImageNet](http://image-net.org/download-images) dataset (if you need to test or train on ImageNet). You may follow the instruction at [fb.resnet.torch](https://github.com/facebook/fb.resnet.torch) to process the validation set.
### Testing
1. Edit `test.sh`. Modify `model` and `data_path` to yours.
Options for `model`: resnet50_ibn_a, resnet50_ibn_b, resnet101_ibn_a, resnext101_ibn_a, se_resnet101_ibn_a, densenet121_ibn_a, densenet169_ibn_a.
2. Run test script
```Shell
sh test.sh
```
### Training
1. Edit `train.sh`. Modify `model` and `data_path` to yours.
2. Run train script
```Shell
sh train.sh
```
### Acknowledgement
This code is developed based on [bearpaw/pytorch-classification](https://github.com/bearpaw/pytorch-classification).
### MXNet Implementation
https://github.com/bruinxiong/IBN-Net.mxnet
### Citing IBN-Net
```
@inproceedings{pan2018IBN-Net,
author = {Xingang Pan, Ping Luo, Jianping Shi, and Xiaoou Tang},
title = {Two at Once: Enhancing Learning and Generalization Capacities via IBN-Net},
booktitle = {ECCV},
year = {2018}
}
```