CompNet

CompNet: Competitive Neural Network for Palmprint Recognition Using Learnable Gabor Kernels

1. Related Materials

Supplementary Material: pdf
Pretrained Models: @
Publicly Available Datasets: Tongji, IITD, REST,NTU, XJTU-UP

2. Visualizations

CB17x17_Gabor feature maps obtained at different epoch

CB17x17_SCC feature maps obtained at different epoch

CB7x7_Conv1 feature maps obtained at different epoch

CB17x17_Conv2 feature maps obtained at different epoch

Each row represents feature maps obtained from one ROI image, and each column corresponds to a single channel.

3. Citation

If it helps you, we would like you to cite the following paper:

@article{spl2021compnet,  
author={Liang, Xu and Yang, Jinyang and Lu, Guangming and Zhang, David},  
journal={IEEE Signal Processing Letters},   
title={CompNet: Competitive Neural Network for Palmprint Recognition Using Learnable Gabor Kernels},   
year={2021},  
volume={28},  
number={},  
pages={1739-1743},  
doi={10.1109/LSP.2021.3103475}}

X. Liang, J. Yang, G. Lu and D. Zhang, "CompNet: Competitive Neural Network for Palmprint Recognition Using Learnable Gabor Kernels," in IEEE Signal Processing Letters, vol. 28, pp. 1739-1743, 2021, doi: 10.1109/LSP.2021.3103475.

4. PyTorch Implementation

Requirements

pytorch-1.2.0
torchvision-0.4.0
cuda 10.2
cudnn 7.6.5.32
python-3.7.4
anaconda-4.9.0
opencv-3.2.7
numpy-1.16.4
sklearn-0.21.3
scipy-1.3.1

Configurations

modify path1 and path2 in genText.py
- path1: path of the training set (e.g., Tongji session1)
- path2: path of the testing set (e.g., Tongji session2)
modify num_classes in train.py and test.py
- Tongji: 600, IITD: 460, REST: 358, XJTU-UP: 200, KTU: 145
modify python_path in train.py and test.py according to which python you are using. ('python')

Commands

cd path/to/CompNet/

#in the CompNet folder:

#generate the training and testing data sets
python ./data/genText.py
mv ./train.txt ./data/
mv ./test.txt ./data/

#train the network
python train.py

#test the model
python test.py

#inference
python inference.py

#Metrics
#obtain the genuine-impostor matching score distribution curve
python    getGI.py   ./rst/veriEER/scores_xxx.txt    scores_xxx

#obtain the EER and the ROC curve
python    getEER.py   ./rst/veriEER/scores_xxx.txt    scores_xxx

The .pth file will be generated at the current folder, and all the other results will be generated in the ./rst folder.

5. Framework

compnet(
  (cb1): CompetitiveBlock(
    (gabor_conv2d): GaborConv2d()
    (argmax): Softmax(dim=1)
    (conv1): Conv2d(9, 32, kernel_size=(5, 5), stride=(1, 1))
    (maxpool): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
    (conv2): Conv2d(32, 12, kernel_size=(1, 1), stride=(1, 1))
  )
  (cb2): CompetitiveBlock(
    (gabor_conv2d): GaborConv2d()
    (argmax): Softmax(dim=1)
    (conv1): Conv2d(9, 32, kernel_size=(5, 5), stride=(1, 1))
    (maxpool): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
    (conv2): Conv2d(32, 12, kernel_size=(1, 1), stride=(1, 1))
  )
  (cb3): CompetitiveBlock(
    (gabor_conv2d): GaborConv2d()
    (argmax): Softmax(dim=1)
    (conv1): Conv2d(9, 32, kernel_size=(5, 5), stride=(1, 1))
    (maxpool): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
    (conv2): Conv2d(32, 12, kernel_size=(1, 1), stride=(1, 1))
  )
  (fc): Linear(in_features=9708, out_features=512, bias=True)
  (drop): Dropout(p=0.25, inplace=False)
  (arclayer): ArcMarginProduct()
)

Acknowledgement

Portions of the research use the REST'2016 Database collected by the Research Groups in Intelligent Machines, University of Sfax, Tunisia. We would also like to thank the organizers (IITD, Tongji, REgim, XJTU, and NTU) for allowing us to use their datasets.

Other References

[1] J. Deng, J. Guo, N. Xue and S. Zafeiriou, “ArcFace: Additive angularmargin loss for deep face recognition,” in Proc. IEEE/CVF Conf.Comput. Vis. Pattern Recognit., Jun. 2019, pp. 4690–4699.

[2] Arcface PyTorch Implementation (ArcMarginProduct). [Online]. Available: https://github.com/ronghuaiyang/arcface-pytorch

[3] P. Chen, W. Li, L. Sun, X. Ning, L. Yu, and L. Zhang, “LGCN: LearnableGabor convolution network for human gender recognition in the wild,” IEICE Trans. Inf. Syst., vol. E102D no. 10, pp. 2067–2071, Oct. 2019.

[4] A. Genovese, V. Piuri, K. N. Plataniotis and F. Scotti, “PalmNet: Gabor-PCA convolutional networks for touchless palmprint recognition,” IEEE Trans. Inf. Forensics Secur., vol. 14, no. 12, pp. 3160–3174, Dec. 2019. doi: 10.1109/TIFS.2019.2911165. PalmNet

JonnyLewis/compnet