Metadata-Version: 1.0
Name: pyeer
Version: 0.3.2
Summary: A python package for biometric and binary classification systems performance evaluation
Home-page: https://www.researchgate.net/profile/Manuel_Aguado_Martinez2
Author: Manuel Aguado Martínez
Author-email: manuelaguadomtz@gmail.com
License: UNKNOWN
Description: =====
        PyEER
        =====
        
        **PyEER** is a python package for biometric systems performance evaluation. Includes ROC, DET, FNMR and FMR curves
        plotting, scores distribution plotting, EER and operating points estimation. It can be also used to evaluate binary
        classification systems.
        
        The program provided within this package receive two files holding genuine match scores and impostor match scores [1].
        Genuine match scores are obtained by matching feature sets of the same class (same person), while impostor matching
        scores are obtained by matching feature sets of different classes (different persons). Using this scores the program 
        plots ROC, DET, FNMR(t), FMR(t) curves and estimates Equal Error Rate Value and operating points for each system. EER values are reported as specified in [2]
        
        Utilities provided within this package can also be used to develop other scripts by importing the module **pyeer**.
        
        **PyEER** has been developed with the idea of providing researchers and the scientific community in general with a 
        tool to correctly evaluate and report the performance of their systems.
        
        Installing
        ==========
        
        .. code:: sh
        
            pip install pyeer
        
        Input file formats
        ==================
        Genuine match scores must be provided in a file with one score per line. Each line can have any number of columns but
        the scores must be in the last column. For impostor match scores the program can handle two different formats:
        
        Histogram format
        ----------------
        
        Although the vast majority of the systems report scores normalized between 0 and 1, there are some that report
        integer scores [3]. When computing a lot of impostor scores, millions of them, it can be computationally 
        expensive to read all those scores from a file. Therefore, in those cases may be worth it to use this format.
        
        **Restrictions:** Only integer scores are supported
        
        **File format:** Each line contains the number of scores equals to the index of the line in the file
        (starting from zero). For example, given a file:
        
        | 123
        | 12
        | 212
        | 321
        | ...
        | ...
        | ...
        |
        
        The above file example indicates that there are 123 scores equals to 0, 12 scores equals to 1, 212 scores
        equals to 2, 321 scores equals to 3 and so on.
        
        **Recommendations:** Use this format for very large experiments (millions of scores).
        
        **Note:** Only impostor scores file can mimic this format.
        
        Non-Histogram format
        --------------------
        
        **Restrictions:** None. Integer and float scores are both supported.
        
        **File format:** All the scores one by line, just as the genuine match scores file format
        
        Usage
        =====
        
        **console cmd:** geteerinf
        
        Examples:
        ---------
        
        
        **To print the script help**
        
        .. code:: sh
        
            geteerinf -h
        
        **One experiment (Non-histogram format):**
        
        .. code:: sh
        
            geteerinf -p "example_files/non_hist/" -i "exp3_false.txt" -g "exp3_true.txt" -e "exp3"
        
        **More than one experiment (Non-histogram format):**
        
        .. code:: sh
        
            geteerinf -p "example_files/non_hist/" -i "exp1_false.txt,exp2_false.txt" -g "exp1_true.txt,exp2_true.txt" -e "exp1,exp2"
        
        **One experiment (Histogram format):**
        
        .. code:: sh
        
            geteerinf -p "example_files/hist/" -i "exp1_false.txt" -g "exp1_true.txt" -e "exp1" -ht
        
        For all the above examples a CSV file will be generated in the directory where the program was invoked. The file contains
        a summary with the values of EER, operating points and area under the ROC curve for each experiment. To specify the 
        directory where to saved it, you can use the "-sp" option.
        
        **Note:** To run the above examples you can download the score files from the project site
        on Gitlab or extract them from inside the package installation
        
        Contributing
        ============
        
        Do you find **PyEER** useful? You can collaborate with us:
        
        `Link Gitlab <https://gitlab.com/manuelaguadomtz/pyeer>`_
        
        References
        ==========
        
        [1] D. Maltoni et al., Handbook of Fingerprint Recognition, Springer-Verlag London Limited 2009
        
        [2] Maio D., Maltoni D., Cappelli R., Wayman J.L. and Jain A.K., “FVC2000: Fingerprint verification
        competition,” IEEE Transactions on Pattern Analysis Machine Intelligence, vol. 24, no. 3, pp. 402–412, 2002
        
        [3] Hernandez-Palancar, J., Munoz-Briseno, A., & Gago-Alonso, A. (2014). Using a
        triangular matching approach for latent fingerprint and palmprint identifica-
        tion. International Journal of Pattern Recognition and Artificial Intelligence, 28, 1460004.
Keywords: Equal Error Rate,False Match Rate,ROC,DET,False Non-Match Rate,EER,FMR,FNMR,ZeroFNMR,ZeroFMR,Biometric Systems
Platform: UNKNOWN
