BEAT - tpereira/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 3

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: probes

Endpoint Name	Data Format	Nature
probe_isv_offset	system/array_1d_floats/1	Input
probe_statistics	tutorial/gmm_statistics/1	Input
template_ids	system/array_1d_uint64/1	Input
probe_id	system/uint64/1	Input
probe_client_id	system/uint64/1	Input
scores	tutorial/probe_scores/1	Output

Group: templates

Endpoint Name	Data Format	Nature
template_client_id	system/uint64/1	Input
template_id	system/uint64/1	Input
template_model	tpereira/isvmachine/1	Input

Group: train

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

import bob
import numpy

def gmm_from_data(data):
    """Unmangles a bob.machine.GMMMachine from a BEAT Data object"""

dim_c, dim_d = data.means.shape
    gmm = bob.machine.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):
    """Unmangles a bob.machine.ISVBase from a BEAT Data object"""

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

def stats_from_data(data):
    """Unmangles a bob.machine.GMMStats from a BEAT Data object"""

dim_c, dim_d = data.sum_px.shape
    stat = bob.machine.GMMStats(dim_c, dim_d)
    stat.t = long(data.t)
    stat.n = data.n
    stat.sum_px = data.sum_px
    stat.sum_pxx = data.sum_pxx
    return stat

def isvmachine_from_data(data, isvbase):
    """Unmangles a bob.machine.ISVBase from a BEAT Data object"""

isvmachine = bob.machine.ISVMachine(isvbase)
    isvmachine.z = data.latent_z
    return isvmachine

class Algorithm:

def __init__(self):
        self.ubm       = None
        self.isvbase   = None
        self.templates = None

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)

# retrieve all the templates once
        if self.templates is None:
            self.templates = {}
            group = inputs.groupOf('template_model')

while group.hasMoreData():
                group.next()

template_id = group['template_id'].data.value

self.templates[template_id] = dict(
                    client_id = group['template_client_id'].data.value,
                    model = isvmachine_from_data(group['template_model'].data, self.isvbase),
                )

# process the probe
        template_ids = inputs['template_ids'].data.value
        statistics = stats_from_data(inputs['probe_statistics'].data)
        ux = inputs['probe_isv_offset'].data.value

scores = []
        for template_id in template_ids:
            template_client_identity = self.templates[template_id]['client_id']
            model = self.templates[template_id]['model']
            
            
            score = model.forward_ux(statistics, ux)

scores.append({
                'template_identity': template_client_identity,
                'score': score,
            })

outputs['scores'].write({
                'client_identity': inputs['probe_client_id'].data.value,
                'scores': scores
            }            
        )

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

Computes the ISV scoring.

Specific details can be found in Equation (40) in [McCool2013].

This algorithm relies on the Bob library.

The inputs are:

probe_isv_offset: The Session Offset of a probe.
probe_statistics: A set of GMM Statistics of a probe.
template_ids: A set of probe (class/subject) identifier as an unsigned 64 bits integer.
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 an unsigned 64 bits integer.
template_model: The client model is the latent variable z_i ( Eq. (31) in [McCool2013]) that corresponds to the client offset (with the session variations suppressed).

The output are the scores.

[McCool2013]

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

Experiments

Attestation:

Privacy:

Status:

			Updated	Name	Databases/Protocols	Analyzers
				smarcel/tpereira/full_isv_multi/2/btas2015_face-periocular_mobio-female_det	mobio/1@female	tutorial/eerhter_postperf_iso/1
				tpereira/tpereira/full_isv/2/btas2015_periocular_mobio-female_det_bobv2-0	mobio/1@female	tutorial/eerhter_postperf_iso/1

Scientific Python 2.7 (0.0.4) 36

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 tpereira isv_scoring 4

Endpoint Groups 3

Group: probes

Group: templates

Group: train

Experiments

algorithms

tpereira

isv_scoring

4