BEAT - siebenkopf/GaborGraphSimilarity/12

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: probes

Endpoint Name	Data Format	Nature
comparison_ids	system/array_1d_uint64/1	Input
probe	siebenkopf/graph/1	Input
scores	system/array_1d_floats/1	Output

Group: models

Endpoint Name	Data Format	Nature
model	siebenkopf/graph_model/3	Input

Parameters 1

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

Name	Description	Type	Default	Range/Choices
gabor_jet_similarity	The Gabor jet similarity function to be used	string	Canberra	ScalarProduct, Canberra, Disparity, PhaseDiff, PhaseDiffPlusCanberra

import bob
import numpy

class Algorithm:

def __init__(self):
    self.models = None
    
  def setup(self, parameters):
    self.similarity = bob.machine.GaborJetSimilarity({
      'ScalarProduct' : bob.machine.gabor_jet_similarity_type.SCALAR_PRODUCT,
      'Canberra' : bob.machine.gabor_jet_similarity_type.CANBERRA,
      'Disparity' : bob.machine.gabor_jet_similarity_type.DISPARITY,
      'PhaseDiff' : bob.machine.gabor_jet_similarity_type.PHASE_DIFF,
      'PhaseDiffPlusCanberra' : bob.machine.gabor_jet_similarity_type.PHASE_DIFF_PLUS_CANBERRA,
    } [parameters.get('gabor_jet_similarity', "Canberra")])
  
    return True
       
  def process(self, inputs, outputs):
    # retrieve all the models once
    if self.models is None:
      self.models = {}
      input = inputs['model']
      while input.hasMoreData(): 
        input.next()
        self.models[input.data.id] = (
           input.data.graphs
        )

# process the probe
    comparison_ids = inputs['comparison_ids'].data.value
    probe = inputs['probe'].data
    probe_jets = [numpy.array([probe.jet[j].abs, probe.jet[j].phase]) for j in range(len(probe.jet))]

scores = []
    for comparison_id in comparison_ids:
      graphs = self.models[comparison_id]
      similarities = [[] for j in range(len(probe.jet))]
      # compute similarities between model graphs and probe graph
      for graph in graphs:
        for j in range(len(graph.jet)):
           model_jet = numpy.array([graph.jet[j].abs, graph.jet[j].phase])
           similarities[j].append(self.similarity(model_jet, probe_jets[j]))
      # for each jet location, get the maximum similarity, and average those
      scores.append(sum([max(sims) for sims in similarities]) / len(probe.jet))

# finally, write all scores
    outputs['scores'].write(
      {
        'value': 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

In this algorithm, the similarity between a template graph T (which is a concatenation of several enrollment graphs) and a probe sample P is computed. The similarities of all node positions n is simply averaged:

sim_total(T, P) = (1)/(N)⎲⎳_nsim_node(T_n, P_n)

In each node, the similarity of all enrollment jets t_m with the probe jet p is computed, and the maximum value is taken:

sim_node(t, p) = max_mS(t_m, p)

Where S is a Gabor jet similarity function, which can be chosen accordingg to [Guenther12].

[Guenther12]

Manuel Günther, Denis Haufe, Rolf P. Würtz. Face recognition with disparity corrected Gabor phase differences. Artificial Neural Networks and Machine Learning, pp. 411-418, 2012.

Experiments

Attestation:

Privacy:

Status:

Name	Databases/Protocols	Analyzers
siebenkopf/siebenkopf/FaceRec-WithOut-Training/2/XM2VTS-PhaseDiff	xm2vts/1@darkened-lp1	siebenkopf/ROC/15,siebenkopf/EER_HTER/8
siebenkopf/siebenkopf/FaceRec-WithOut-Training/2/XM2VTS-ScalarProduct	xm2vts/1@darkened-lp1	siebenkopf/ROC/15,siebenkopf/EER_HTER/8
siebenkopf/siebenkopf/FaceRec-WithOut-Training/2/XM2VTS-Canberra	xm2vts/1@darkened-lp1	siebenkopf/ROC/15,siebenkopf/EER_HTER/8
siebenkopf/siebenkopf/FaceRec-WithOut-Training/2/Banca_P-ScalarProduct	banca/1@P	siebenkopf/ROC/15,siebenkopf/EER_HTER/8
siebenkopf/siebenkopf/FaceRec-WithOut-Training/2/Banca_P-Canberra	banca/1@P	siebenkopf/ROC/14,siebenkopf/EER_HTER/8
siebenkopf/siebenkopf/FaceRec-WithOut-Training/2/Banca_P-PhaseDiff	banca/1@P	siebenkopf/ROC/14,siebenkopf/EER_HTER/8

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 siebenkopf GaborGraphSimilarity 12

Endpoint Groups 2

Group: probes

Group: models

Parameters 1

Experiments

algorithms

siebenkopf

GaborGraphSimilarity

12