Advancing 3D Finger Knuckle Recognition via Deep Feature Learning [1]

This package provides the implementation codes for reproducing the experimental results presented in this paper [1]. This package also re-implements the baseline method presented in reference [2].

This package is tested using a computer system: with Ubuntu 18.04.6 LTS and Pytorch 1.11.0.

Detail steps:

1. Train/Test baseline methods and the FKNet+. -Execute main.ipynb Jupyter notebook.

2. Visualize the recognition performances. -Execute matlab/ROC_model_select.m using MATLAB.

References

• [1] Kevin H. M. Cheng, Xu Cheng, and Guoying Zhao. Advancing 3D Finger Knuckle Recognition via Deep Feature Learning. arXiv preprint arXiv:2301.02934.
• [2] Kevin H. M. Cheng, and Ajay Kumar. Deep Feature Collaboration for Challenging 3D Finger Knuckle Recognition. IEEE Transactions on Information Forensics and Security (TIFS), 16, pp.1158-1173, 2021.