BEAT - pkorshunov/isv

This algorithm is a legacy one. The API has changed since its implementation. New versions and forks will need to be updated.

This algorithm is splittable

Endpoint Groups 2

Algorithms have at least one input and one output. All algorithm endpoints are organized in groups. Groups are used by the platform to indicate which inputs and outputs are synchronized together. The first group is automatically synchronized with the channel defined by the block in which the algorithm is deployed.

Group: templates

Endpoint Name	Data Format	Nature
statistics	tutorial/gmm_statistics/1	Input
template_id	system/text/1	Input
model	tpereira/isvmachine/1	Output

Group: train

Endpoint Name	Data Format	Nature
ubm	tutorial/gmm/1	Input
isvbase	tpereira/isvbase/1	Input

Parameters 1

Parameters allow users to change the configuration of an algorithm when scheduling an experiment

Name	Description	Type	Default	Range/Choices
isv-enroll-iterations		uint32	1

import bob.learn.em
import numpy

def gmm_from_data(data):
    """Reads a bob.learn.em.GMMMachine from a BEAT Data object"""

dim_c, dim_d = data.means.shape
    gmm = bob.learn.em.GMMMachine(dim_c, dim_d)
    gmm.weights = data.weights
    gmm.means = data.means
    gmm.variances = data.variances
    gmm.variance_thresholds = data.variance_thresholds
    return gmm

def isvbase_from_data(data, ubm):
    """Reads a bob.learn.em.ISVBase from a BEAT Data object"""

dim_cd, dim_u = data.subspace_u.shape
    isvbase = bob.learn.em.ISVBase(ubm, dim_u)
    isvbase.u = data.subspace_u
    isvbase.d = data.subspace_d
    return isvbase

def stats_from_data(data):
    """Reads a bob.learn.em.GMMStats from a BEAT Data object"""

dim_c, dim_d = data.sum_px.shape
    stat = bob.learn.em.GMMStats(dim_c, dim_d)
    stat.t = int(data.t)
    stat.n = data.n
    stat.sum_px = data.sum_px
    stat.sum_pxx = data.sum_pxx
    return stat

class Algorithm:

def __init__(self):
        self.ubm      = None
        self.isvbase  = None
        self.statistics = []
        self.isv_enroll_iterations = 1

def setup(self, parameters):
        self.isv_enroll_iterations = parameters.get('isv-enroll-iterations', self.isv_enroll_iterations) 
        return True

def process(self, inputs, outputs):
        # retrieve the UBM once
        if self.ubm is None:
            inputs['ubm'].next()
            self.ubm = gmm_from_data(inputs['ubm'].data)

# retrieve the ISVBase once
        if self.isvbase is None:
            inputs['isvbase'].next()
            self.isvbase = isvbase_from_data(inputs['isvbase'].data, self.ubm)

# collect all the features for the current template
        self.statistics.append(stats_from_data(inputs["statistics"].data))

# adapts the UBM GMM for the template (when all the features have been collected)
        if inputs["template_id"].isDataUnitDone():
            model = bob.learn.em.ISVMachine(self.isvbase)

trainer = bob.learn.em.ISVTrainer()
            trainer.enroll(model, self.statistics, int(self.isv_enroll_iterations))

# outputs data
            outputs["model"].write({
                'latent_z':             model.z,
            })

self.statistics = []

return True

The code for this algorithm in Python
The ruler at 80 columns indicate suggested POSIX line breaks (for readability).
The editor will automatically enlarge to accomodate the entirety of your input
Use keyboard shortcuts for search/replace and faster editing. For example, use Ctrl-F (PC) or Cmd-F (Mac) to search through this box

Given a feature vector, a GMM and a U subspace, computes the Intersession Variability Modeling (ISV) client model. Basically, this algorithm computes the latent variable z_i excluding possible session factors (described by the latent variable x_i, j).

Specific details can be found in [McCool2013]:

This algorithm relies on the Bob library.

The inputs are:

statistics: A set of GMM Statistics of a client for enrollment.
ubm: A GMM corresponding to the Universal Background Model.
isvbase: The subspace_u and subspace_d for the session and the client offset respectivelly.
template_id: Client (class/subject) identifier as a text string.

The output, model, is the latent variable z_i (Eq. (31) in McCool2013) that corresponds to the client offset (with the session variations suppressed).

[McCool2013]

McCool, et al.: Session variability modelling for face authentication. IET biometrics 2.3 (2013): 117-129.

Experiments

Attestation:

Privacy:

Status:

Name	Databases/Protocols	Analyzers
pkorshunov/pkorshunov/isv-asv-pad-fusion-complete/1/asv_isv-pad_lbp_hist_ratios_lr-fusion_lr-pa_aligned	avspoof/2@physicalaccess_verify_train,avspoof/2@physicalaccess_verification,avspoof/2@physicalaccess_verification_spoof,avspoof/2@physicalaccess_verify_train_spoof,avspoof/2@physicalaccess_antispoofing	pkorshunov/spoof-score-fusion-roc_hist/1
pkorshunov/pkorshunov/isv-asv-pad-fusion-complete/1/asv_isv-pad_gmm-fusion_lr-pa	avspoof/2@physicalaccess_verify_train,avspoof/2@physicalaccess_verification,avspoof/2@physicalaccess_verification_spoof,avspoof/2@physicalaccess_verify_train_spoof,avspoof/2@physicalaccess_antispoofing	pkorshunov/spoof-score-fusion-roc_hist/1
pkorshunov/pkorshunov/isv-speaker-verification-spoof/1/isv-speaker-verification-spoof-pa	avspoof/2@physicalaccess_verification_spoof,avspoof/2@physicalaccess_verification	pkorshunov/eerhter_postperf_iso_spoof/1
pkorshunov/pkorshunov/isv-speaker-verification/1/isv-speaker-verification-licit	avspoof/2@physicalaccess_verification	pkorshunov/eerhter_postperf_iso/1

Scientific Python 2.7 (0.1.0) 70

This table shows the number of times this algorithm has been successfully run using the given environment. Note this does not provide sufficient information to evaluate if the algorithm will run when submitted to different conditions.

algorithms pkorshunov isv_enroller 2

Endpoint Groups 2

Group: templates

Group: train

Parameters 1

Experiments

algorithms

pkorshunov

isv_enroller

2