TripleE-DG

KO | EN

gitlite — search

#javascript #python #react #ai #llm #linux #macos #nodejs #vue #awesome-list #pytorch #mcp #ios #openai

(★ 10)

No description available.

TripleE-DG Latest Version Download

Download Latest Version (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

## TripleE: Easy Domain Generalization via Episodic Replay

By [Xiaomeng Li](https://xmengli.github.io/), [Hongyu Ren](http://hyren.me/), [Huifeng Yao](https://scholar.google.com/citations?hl=en&user=hzNA39EAAAAJ) and [Ziwei Liu](https://liuziwei7.github.io/).

This repository is for our paper [TripleE: Easy Domain Generalization via Episodic Replay](https://arxiv.org/pdf/2210.01807.pdf).

## Installation

```
conda env create -f IDC_environment.yml
```

## Data Preparation
* Download [PACS dataset](https://drive.google.com/drive/folders/1SKvzI8bCqW9bcoNLNCrTGbg7gBSw97qO), 
[Digits-DG dataset](https://drive.google.com/uc?id=15V7EsHfCcfbKgsDmzQKj_DfXt_XYp_P7), 
[Office-Home dataset](https://drive.google.com/file/d/0B81rNlvomiwed0V1YUxQdC1uOTg/view?resourcekey=0-2SNWq0CDAuWOBRRBL7ZZsw),
[VLCS dataset](https://drive.google.com/uc?id=1skwblH1_okBwxWxmRsp9_qi15hyPpxg8),
[TerraInc dataset](https://lilablobssc.blob.core.windows.net/caltechcameratraps/eccv_18_all_images_sm.tar.gz),
[DomainNet dataset](http://ai.bu.edu/M3SDA/).
You can download these datasets using this [script](https://github.com/facebookresearch/DomainBed/blob/main/domainbed/scripts/download.py).

* Put data under `./DATASET/`
* The correct path should be 
```
./DATASET/PACS/kfold/sketch/...
./DATASET/digits_dg/mnist/...
./DATASET/OfficeHome/Art/...  
./DATASET/VLCS/Caltech101...
./DATASET/terra_incognita/L38...
```

## Training
* Check training command in `./scripts/`. 
* Change `times` to `1,2,3` to run the experiment for 3 times. 
* Set `target` to get results on different target domains.

* By averaging three results on each domain, you will get the following results:

| Sketch    | Cartoon   | Art    |  Photo  | Average 
| ---------- | :-----------:  | :-----------: | :-----------: | :-----------:
| 84.45%    | 80.84%   | 85.30%     | 96.27% | 86.72%

<!-- * Evaluate our models: download models from [cartoon](https://gohkust-my.sharepoint.com/:u:/g/personal/eexmli_ust_hk/EQUZU9JnCV5Hje30gTN29BkBNDiFYmBQXbDa1F2Gypn59g?e=j56u9G), [art_painting](https://gohkust-my.sharepoint.com/:u:/g/personal/eexmli_ust_hk/EXgBELtc0VFNpgDQnMeufxEB_EkAfTbF5-pvzQrSEfcNbA?e=CnRCKG), [photo](https://gohkust-my.sharepoint.com/:u:/g/personal/eexmli_ust_hk/Ef-B6X98bBNPtH3v2qKLVo0BKE7GlyNSkaWM0T91abOoLg?e=5o8fbG), [sketch](https://gohkust-my.sharepoint.com/:u:/g/personal/eexmli_ust_hk/ETlPaYXHTUdFkc82wkrL5y4BUs7Oom8exsrg-clk0zuyRg?e=ZvvDQU). 
Specify the path `--resume` for models and run 
```
python main_dg.py -a resnet18 --lr 0.001 --epochs 100 --batch-size 4  .  --source cartoon  sketch   photo --target art_painting   --ratio 1.0  --result exp/dg/dg_art  --baug 4 --gpu 0 --times 1  --ncesole  --evaluate --resume exp/model_best_art.pth.tar 
```

-->
## Citation

If this code is useful for your research, please consider citing:

<!-- 
  ```shell
@article{li2020self,
  title={Self-supervised Feature Learning via Exploiting Multi-modal Data for Retinal Disease Diagnosis},
  author={Li, Xiaomeng and Jia, Mengyu and Islam, Md Tauhidul and Yu, Lequan and Xing, Lei},
  journal={IEEE Transactions on Medical Imaging},
  year={2020},
  publisher={IEEE}
}

``` -->

TripleE-DG

TripleE-DG Latest Version Download

Today's picks

Was this content helpful?

Recent Feedback

TripleE-DG

TripleE-DG Latest Version Download

Today's picks

Was this content helpful?

Recent Feedback

Download README