def write_biometric_reference(self, model: GMMMachine, model_file):
"""Write the enrolled reference (MAP GMMMachine) into a file."""
return model.save(model_file)
def test_GMMMachine():
# Test a GMMMachine basic features
weights = np.array([0.5, 0.5], "float64")
weights2 = np.array([0.6, 0.4], "float64")
means = np.array([[3, 70, 0], [4, 72, 0]], "float64")
means2 = np.array([[3, 7, 0], [4, 72, 0]], "float64")
variances = np.array([[1, 10, 1], [2, 5, 2]], "float64")
variances2 = np.array([[10, 10, 1], [2, 5, 2]], "float64")
varianceThresholds = np.array([[0, 0, 0], [0, 0, 0]], "float64")
varianceThresholds2 = np.array([[0.0005, 0.0005, 0.0005], [0, 0, 0]],
"float64")
# Initializes a GMMMachine
gmm = GMMMachine(n_gaussians=2)
# Sets the weights, means, variances and varianceThresholds and
# Checks correctness
gmm.weights = weights
gmm.means = means
gmm.variances = variances
gmm.variance_thresholds = varianceThresholds
assert gmm.shape == (2, 3)
np.testing.assert_equal(gmm.weights, weights)
np.testing.assert_equal(gmm.means, means)
np.testing.assert_equal(gmm.variances, variances)
np.testing.assert_equal(gmm.variance_thresholds, varianceThresholds)
newMeans = np.array([[3, 70, 2], [4, 72, 2]], "float64")
newVariances = np.array([[1, 1, 1], [2, 2, 2]], "float64")
# Checks particular varianceThresholds-related methods
varianceThresholds1D = np.array([0.3, 1, 0.5], "float64")
gmm.variance_thresholds = varianceThresholds1D
np.testing.assert_equal(gmm.variance_thresholds, varianceThresholds1D)
gmm.variance_thresholds = 0.005
np.testing.assert_equal(gmm.variance_thresholds, 0.005)
gmm.means = newMeans
gmm.variances = newVariances
np.testing.assert_equal(gmm.means, newMeans)
np.testing.assert_equal(gmm.variances, newVariances)
# Checks comparison
gmm2 = deepcopy(gmm)
gmm3 = GMMMachine(n_gaussians=2)
gmm3.weights = weights2
gmm3.means = means
gmm3.variances = variances
gmm3.variance_thresholds = varianceThresholds
gmm4 = GMMMachine(n_gaussians=2)
gmm4.weights = weights
gmm4.means = means2
gmm4.variances = variances
gmm4.variance_thresholds = varianceThresholds
gmm5 = GMMMachine(n_gaussians=2)
gmm5.weights = weights
gmm5.means = means
gmm5.variances = variances2
gmm5.variance_thresholds = varianceThresholds
gmm6 = GMMMachine(n_gaussians=2)
gmm6.weights = weights
gmm6.means = means
gmm6.variances = variances
gmm6.variance_thresholds = varianceThresholds2
assert_gmm_equal(gmm, gmm2)
assert (gmm != gmm2) is False
assert gmm.is_similar_to(gmm2)
assert gmm != gmm3
assert gmm.is_similar_to(gmm3) is False
assert gmm != gmm4
assert gmm.is_similar_to(gmm4) is False
assert gmm != gmm5
assert gmm.is_similar_to(gmm5) is False
assert gmm != gmm6
assert gmm.is_similar_to(gmm6) is False
# Saving and loading
with tempfile.NamedTemporaryFile(suffix=".hdf5") as f:
filename = f.name
gmm.save(HDF5File(filename, "w"))
# Using from_hdf5
gmm1 = GMMMachine.from_hdf5(HDF5File(filename, "r"))
assert type(gmm1.n_gaussians) is np.int64
assert type(gmm1.update_means) is np.bool_
assert type(gmm1.update_variances) is np.bool_
assert type(gmm1.update_weights) is np.bool_
assert type(gmm1.trainer) is str
assert gmm1.ubm is None
assert_gmm_equal(gmm, gmm1)
# Using load
gmm1 = GMMMachine(n_gaussians=gmm.n_gaussians)
gmm1.load(HDF5File(filename, "r"))
assert type(gmm1.n_gaussians) is np.int64
assert type(gmm1.update_means) is np.bool_
assert type(gmm1.update_variances) is np.bool_
assert type(gmm1.update_weights) is np.bool_
assert type(gmm1.trainer) is str
assert gmm1.ubm is None
assert_gmm_equal(gmm, gmm1)
with tempfile.NamedTemporaryFile(suffix=".hdf5") as f:
filename = f.name
gmm.save(filename)
gmm1 = GMMMachine.from_hdf5(filename)
assert_gmm_equal(gmm, gmm1)
# Weights
n_gaussians = 5
machine = GMMMachine(n_gaussians)
default_weights = np.full(shape=(n_gaussians, ),
fill_value=1.0 / n_gaussians)
default_log_weights = np.full(shape=(n_gaussians, ),
fill_value=np.log(1.0 / n_gaussians))
# Test weights getting and setting
np.testing.assert_almost_equal(machine.weights, default_weights)
np.testing.assert_almost_equal(machine.log_weights, default_log_weights)
modified_weights = default_weights
modified_weights[:n_gaussians // 2] = (1 / n_gaussians) / 2
modified_weights[n_gaussians // 2 +
n_gaussians % 2:] = (1 / n_gaussians) * 1.5
# Ensure setter works (log_weights is updated correctly)
machine.weights = modified_weights
np.testing.assert_almost_equal(machine.weights, modified_weights)
np.testing.assert_almost_equal(machine.log_weights,
np.log(modified_weights))
