This folder contains dataset for training and evaluation.

Here are 4 subfolders as below.

original
train_data
real_data
np_data

This folder contains original fingerprint images.

The image size must be greater than 160x160.

The image file must be a 8bit grayscale bitmap(.bmp) file.

And the file name must be "<FingerID (5 digits)>_<Index (2 digits)>.bmp"

For example, if dataset contains 3 fingers and each finger has two images, the folder structure will be like below.

dataset\original
    00000_00.bmp
    00000_01.bmp
    00001_00.bmp
    00001_01.bmp
    00002_00.bmp
    00002_01.bmp

This folder contains fingerprint images for training.

Training images can be generated with "01_01_prepare_train_data.py" file.

This file can generate training images from original images.

python 01_01_prepare_train_data.py

This folder contains clear fingerprint images.

Clear images can be generated with "01_02_prepare_real_data.py" file.

This file will copy first image for each finger from train_data folder.

python 01_02_prepare_real_data.py

This folder contains numpy data for training and evaluation.

Numpy data file can be generated with "01_03_prepare_dataset.py" file.

python 01_03_prepare_dataset.py

You can get sample dataset from here.

Fingerprint Dataset for FVC2000_DB4_B in BaiduYun, Password: xeqp

Fingerprint Dataset for FVC2000_DB4_B in Google Drive

Fingerprint Dataset for FVC2000_DB4_B in Kaggle

This dataset was created from FVC2000_DB4_B.

Please donate fingerprint images for research purpose.

Thanks.