Use numpy.spatial.distance.euclidean instead of lambda function.

faceverify/eigenface.py

@@ -21,7 +21,7 @@
 import bob
 import xbob.db.atnt
 import os, sys
-import numpy, numpy.linalg
+import numpy, scipy.spatial
 from matplotlib import pyplot
 
 # This is the base directory where by default the AT&T images are found. You can
@@ -78,6 +78,8 @@ def extract_feature(image, pca_machine):
   return projected_feature
 
 
+DISTANCE_FUNCTION = scipy.spatial.distance.euclidean
+
 def main():
   """This function will perform an eigenface test on the AT&T database"""
 
@@ -130,13 +132,12 @@ def main():
   negative_scores = []
 
   print "Computing scores"
-  distance_function = (lambda x,y: numpy.linalg.norm(x - y))
 
   # iterate through models and probes and compute scores
   for model_key, model_feature in model_features.iteritems():
     for probe_key, probe_feature in probe_features.iteritems():
       # compute score as the negative Euclidean distance
-      score = - distance_function(model_feature, probe_feature)
+      score = - DISTANCE_FUNCTION(model_feature, probe_feature)
 
       # check if this is a positive score
       if atnt_db.get_client_id_from_file_id(model_key) == atnt_db.get_client_id_from_file_id(probe_key):

faceverify/tests/__init__.py

@@ -55,7 +55,7 @@ class FaceVerifyExampleTest(unittest.TestCase):
   def test01_eigenface(self):
     # test the eigenface algorithm
     try:
-      from faceverify.eigenface import load_images, train, extract_feature
+      from faceverify.eigenface import load_images, train, extract_feature, DISTANCE_FUNCTION
     except ImportError as e:
       raise SkipTest("Skipping the tests since importing from faceverify.eigenface raised exception '%s'"%e)
 
@@ -92,7 +92,7 @@ class FaceVerifyExampleTest(unittest.TestCase):
     self.assertTrue(numpy.allclose(probe_ref, probe))
 
     # compute score
-    score = numpy.linalg.norm(model - probe)
+    score = DISTANCE_FUNCTION(model, probe)
     self.assertAlmostEqual(score, 3498.308154114)