# DeepBDC For Few-shot Larning
## Introduction
In this repo, we provide the implementation of the following paper:
"Joint Distribution Matters: Deep Brownian Distance Covariance for Few-Shot Classification" [[Project]](http://peihuali.org/DeepBDC/index.html) [[Paper]](https://arxiv.org/pdf/2204.04567.pdf).
In this paper, we propose deep Brownian Distance Covariance (DeepBDC) for few-shot classification. DeepBDC can effectively learn image representations by measuring, for the query and support images, the discrepancy between the joint distribution of their embedded features and product of the marginals. The core of DeepBDC is formulated as a modular and efficient layer, which can be flexibly inserted into deep networks, suitable not only for meta-learning framework based on episodic training, but also for the simple transfer learning (STL) framework of pretraining plus linear classifier.
If you find this repo helpful for your research, please consider citing our paper：
```
@inproceedings{DeepBDC-CVPR2022,
title={Joint Distribution Matters: Deep Brownian Distance Covariance for Few-Shot Classification},
author={Jiangtao Xie and Fei Long and Jiaming Lv and Qilong Wang and Peihua Li},
booktitle={CVPR},
year={2022}
}
```
## Few-shot classification Results
Experimental results on miniImageNet, CUB and tieredImageNet. We report average results with 2,000 randomly sampled episodes for both 1-shot and 5-shot evaluation. More details on the experiments can be seen in the paper.
### miniImageNet
*We followed [DeepEMD](https://github.com/icoz69/DeepEMD) for data preprocessing.*
*Note that for Good-Embed and STL DeepBDC, a sequential self-distillation technique is used to obtain the pre-trained models; See the paper of Good-Embed for details.*
### CUB
*We followed [CloserLookFewShot](https://github.com/wyharveychen/CloserLookFewShot) for data preprocessing.*
*Note that for Good-Embed and STL DeepBDC, a sequential self-distillation technique is used to obtain the pre-trained models; See the paper of Good-Embed for details.*
### tieredImageNet
*We followed [DeepEMD](https://github.com/icoz69/DeepEMD) for data preprocessing.*
*Note that for Good-Embed and STL DeepBDC, a sequential self-distillation technique is used to obtain the pre-trained models; See the paper of Good-Embed for details.*
## References
[BDC] G. J. Szekely and M. L. Rizzo. Brownian distance covariance. Annals of Applied Statistics, 3:1236–1265, 2009.
[ProtoNet] Jake Snell, Kevin Swersky, and Richard Zemel. Prototypical networks for few-shot learning. In NIPS, 2017.
[Good-Embed] Y. Tian, Y. Wang, D. Krishnan, J. B. Tenenbaum, and P. Isola. Rethinking few-shot image classification: a good embedding is all you need? In ECCV, 2020.
## Implementation details
### Datasets
- miniImageNet: Download Link: [[BaiduCloud](https://pan.baidu.com/s/1Wi06keM-1WXP26YqwdpaFw?pwd=ankq)] [[GoogleDrive](https://drive.google.com/file/d/1aBxfcU5cn-htIlqriiOQCOXp_t9TOm9g/view?usp=sharing)].
- CUB: Download Link: [[BaiduCloud](https://pan.baidu.com/s/1JyVQC1-cLiPIl6yYAdlkeA?pwd=yrv1)] [[GoogleDrive](https://drive.google.com/file/d/1sbOiZP-U4A7NdhkJo7YzeffNf5GatIwk/view?usp=sharing)].
- tieredImageNet: Download Link: [[BaiduCloud](https://pan.baidu.com/s/14qKXc8982Ioso_0Fpt24QQ?pwd=90w2)]
### Implementation environment
*Note that the test accuracy may slightly vary with different Pytorch/CUDA versions, GPUs, etc.*
- Linux
- Python 3.8.3
- torch 1.7.1
- GPU (RTX3090) + CUDA11.0 CuDNN
- sklearn1.0.1, pillow8.0.0, numpy1.19.2
### Installation
- Clone this repo:
```bash
git clone https://github.com/Fei-Long121/DeepBDC.git
cd DeepBDC
```
### **For Meta DeepBDC on general object recognition**
1. `cd scripts/mini_magenet/run_meta_deepbdc`
2. modify the dataset path in `run_pretrain.sh`, `run_metatrain.sh` and `run_test.sh`
3. `bash run.sh`
### **For STL DeepBDC on general object recognition**
1. `cd scripts/mini_imagenet/run_stl_deepbdc`
2. modify the dataset path in `run_pretrain.sh`, `run_distillation.sh` and `run_test.sh`
3. `bash run.sh`
## Acknowledgments
Our code builds upon the the following code publicly available:
- [CloserLookFewShot](https://github.com/wyharveychen/CloserLookFewShot)
- [RFS](https://github.com/WangYueFt/rfs/)
## Contact
If you have any questions or suggestions, please contact us:
`Fei Long(longfei121@mail.dlut.edu.cn)`
`Jiaming Lv(ljm_vlg@mail.dlut.edu.cn)`
