Towards Protecting and Enhancing Vascular Biometric Recognition methods via Biohashing and Deep Neural Networks

This package is part of the signal-processing and machine learning toolbox Bob. It contains the source code to reproduce the following paper:

@article{TBIOM2021_ProtectVascularBiohashDNN,
  title={Towards Protecting and Enhancing Vascular Biometric Recognition methods via Biohashing and Deep Neural Networks},
  author={Shahreza, Hatef Otroshi and Marcel, S{\'e}bastien},
  journal={IEEE Transactions on Biometrics, Behavior, and Identity Science},
  year={2021},
  publisher={IEEE},
  volume={3},
  number={3},
  pages={394-404},
  doi={10.1109/TBIOM.2021.3076444}
}

If you use this package and/or its results, please cite both the package and the corresponding paper. Also, please ensure that you include the two original Bob references in your citations.

NOTE: In the instructions that follow, the use of < ... > inside file paths indicates that you must replace the part enclosed within < and > with the appropriate value. For example <path_to_your_working_directory> indicates that you should replace this part with the full path to the directory inside which you are running this code.

Installation

The experiments can only be executed on a Linux 64-bit machine. Install conda and run the steps below::

$ git clone https://gitlab.idiap.ch/bob/bob.paper.tbiom2021_protect_vascular_dnn_biohash.git
$ cd bob.paper.tbiom2021_protect_vascular_dnn_biohash
$ conda create -n TBIOM2021_ProtectVacularBiohashDNN --file package-list.txt
$ conda activate TBIOM2021_ProtectVacularBiohashDNN

Downloading the Databases

The next thing you must do is to download the databases used in our experiments:

1. UTFVP Fingervein Dataset: Please refer to the this link to download this database.
2. PUT Vein Dataset: Please refer to the this link to download this database.
3. VeraFinger Dataset: Please refer to the this link to download this database.

Take note of the directory in which you have stored the downloaded database. Then, create a textfile named .bob_bio_databases.txt and store it in your home directory. Inside this textfile, insert the following line:

[YOUR_UTFVP_DIRECTORY] = <path_to_utfvp_database>
[YOUR_PUTVEIN_IMAGE_DIRECTORY] = <path_to_putvein_database>
[YOUR_VERAFINGER_DIRECTORY] = <path_to_verafinger_database>

Make sure you replace <path_to_utfvp_database>, <path_to_putvein_database>, and <path_to_verafinger_database> with the path to your downloaded datasets.

---

Running the Experiments

You are now ready to run the experiments to reproduce the results in the paper. Experiments for each dataset are descibed separately:

1. Experiemnts on UTFVP Dataset:

[UTFVP] Step 1: Preparing data for training (Data Augmentation)

In order to prepare the data for training the neural network you need to run data.py in data directory:

$ cd UTFVP
$ cd data
$ ./task.sh

NOTE: If you are at Idiap, you can use task_idiap.sh to run on the grid.

NOTE: In data directory, there is a Jupyter notebook Sample_Image.ipynb, which contains the code to generate images used in figure 1 of the paper.

[UTFVP] Step 2: Training Deep Auto-Encoders

Now that the data is ready, we can train our DNN. For this end, run the following code:

$ cd UTFVP
$ ./train_dnns.sh

[UTFVP] Step 3: Executing different finger vein recognition methods

We can execute different finger vein recognition methods including WLD, WLD+PCA, , WLD+Biohash, MC, MC+PCA, MC+Biohash, RLT, RLT+PCA, RLT+Biohash and our trained AE+Biohash [the proposed method] by runnung the following script:

$ cd UTFVP
$ ./execute_fvr.sh

[UTFVP] Step 4: Evaluation

The UTFVP/evaluation folder contains bash files and Jupyter Notebooks to generate the plots and values in the tables. The tree of the files in this folder is as below:

├── FVR
│   ├── alpha
│   │   └── eval.sh
│   ├── eval.sh
│   └── TSNE.ipynb
└── Protect_Enhace_prevFVR
    ├── ablation
    │   ├── alpha
    │   │   └── eval.sh
    │   ├── Bhsh_lngth
    │   │   ├── eval.sh
    │   │   └── LowerBand
    │   │       └── eval.sh
    │   └── emb
    │       └── eval.sh
    ├── Exp1_compare
    │   ├── EER.ipynb
    │   ├── eval-100.sh
    └── TSNE
        └── TSNE.ipynb

You can run each of the bash files or Jupyter Notebooks to generate the results.

---

2. Experiemnts on PUT Vein Dataset:

2.1. PUT Vein Wist Dataset

[PUT-wist] Step 1: Preparing data for training (Data Augmentation)

In order to prepare the data for training the neural network you need to use Data.ipynb in data directory. You can also run feats.py file after you generated palm_all_imgs_aug.npy using Data.ipynb:

$ cd PUT
$ cd PUT-wist
$ cd data
$ python feats.py

NOTE: If you are at Idiap, you can use task_idiap.sh to run on the grid after you generated palm_all_imgs_aug.npy using Data.ipynb:

$ cd PUT
$ cd PUT-wist
$ cd data
$ ./task_idiap.sh

[PUT-wist] Step 2: Training Deep Auto-Encoders

Now that the data is ready, we can train our DNN. For this end, run the following code:

$ cd PUT
$ cd PUT-wist
$ ./train_dnns.sh

[PUT-wist] Step 3: Executing different finger vein recognition methods

We can execute different finger vein recognition methods including WLD, WLD+PCA, , WLD+Biohash, MC, MC+PCA, MC+Biohash, RLT, RLT+PCA, RLT+Biohash and our trained AE+Biohash [the proposed method] by runnung the following script:

$ cd PUT
$ cd PUT-wist
$ ./execute_fvr.sh

[PUT-wist] Step 4: Evaluation

You can run the bash file to generate the ROC plot and evaluation metrics:

$ cd PUT
$ cd PUT-wist
$ cd evaluation/Protect_Enhace_prevFVR/Exp1_compare
$ ./eval.sh

---

2.2. PUT Vein Palm Dataset

[PUT-palm] Step 1: Preparing data for training (Data Augmentation)

In order to prepare the data for training the neural network you need to use Data.ipynb in data directory. You can also run feats.py file after you generated palm_all_imgs_aug.npy using Data.ipynb:

$ cd PUT
$ cd PUT-palm
$ cd data
$ python feats.py

NOTE: If you are at Idiap, you can use task_idiap.sh to run on the grid after you generated palm_all_imgs_aug.npy using Data.ipynb:

$ cd PUT
$ cd PUT-palm
$ cd data
$ ./task_idiap.sh

[PUT-palm] Step 2: Training Deep Auto-Encoders

Now that the data is ready, we can train our DNN. For this end, run the following code:

$ cd PUT
$ cd PUT-palm
$ ./train_dnns.sh

[PUT-palm] Step 3: Executing different finger vein recognition methods

We can execute different finger vein recognition methods including WLD, WLD+PCA, , WLD+Biohash, MC, MC+PCA, MC+Biohash, RLT, RLT+PCA, RLT+Biohash and our trained AE+Biohash [the proposed method] by runnung the following script:

$ cd PUT
$ cd PUT-palm
$ ./execute_fvr.sh

[PUT-palm] Step 4: Evaluation

You can run the bash file to generate the ROC plot and evaluation metrics:

$ cd PUT
$ cd PUT-palm
$ cd evaluation/Protect_Enhace_prevFVR/Exp1_compare
$ ./eval.sh

---

3. Experiemnts on VERA Finger Dataset:

[VERAFinger] Step 1: Preparing data for training (Data Augmentation)

In order to prepare the data for training the neural network you need to use Data.ipynb in data directory.

NOTE: If you are at Idiap, you can use task_idiap.sh to use grid to extract WLD, RLT, and MC features after you generated the augmented data using Data.ipynb.

$ cd VERAFinger
$ cd data
$ ./task_idiap.sh

[VERAFinger] Step 2: Training Deep Auto-Encoders

Now that the data is ready, we can train our DNN. For this end, run the following code:

$ cd VERAFinger
$ ./train_dnns.sh

[VERAFinger] Step 3: Executing different finger vein recognition methods

We can execute different finger vein recognition methods including WLD, WLD+PCA, , WLD+Biohash, MC, MC+PCA, MC+Biohash, RLT, RLT+PCA, RLT+Biohash and our trained AE+Biohash [the proposed method] by runnung the following script:

$ cd VERAFinger
$ ./execute_fvr.sh

[VERAFinger] Step 4: Evaluation

You can run the bash file to generate the ROC plot and evaluation metrics:

$ cd VERAFinger/evaluation/Protect_Enhace_prevFVR/Exp1_compare
$ ./eval.sh

Contact

For questions or reporting issues to this software package, contact our development mailing list.