def test_simple_unique_rand_value():
# -- creating a fake y_prediction
y_pred = np.eye(32).astype('f')
y_pred = gaussian_filter(y_pred, 2.)
y_pred -= y_pred.min()
y_pred /= y_pred.max()
y_pred = 1. - y_pred
# -- creating a fake ground truth
y_true = 1 - np.eye(32).astype(np.uint8)
re = rand_error(y_true, y_pred, th_min=0.6, th_max=0.7)
assert np.abs(re - 0.13515) < EPSILON
def test_simple_segmentation_with_provided_path():
# -- creating a fake y_prediction
y_pred = np.eye(32).astype('f')
y_pred = gaussian_filter(y_pred, 2.)
y_pred -= y_pred.min()
y_pred /= y_pred.max()
y_pred = 1. - y_pred
# -- creating a fake ground truth
y_true = 1 - np.eye(32).astype(np.uint8)
pe = pixel_error(y_true, y_pred)
re = rand_error(y_true, y_pred)
we = warp_error(y_true, y_pred)
assert np.abs(pe - 0.010183) < EPSILON
assert np.abs(re - 0.020385) < EPSILON
assert np.abs(we - 0.0) < EPSILON
