def test_quickshift_medoid_maps():
i = img.copy()
maps, gaps, estimate = quickshift(i,
kernel_size=2,
max_dist=10,
medoid=True)
assert maps.shape == (512, 512)
assert_allclose(maps[0:5, 0], [514., 1026., 1026., 1538., 1538.],
rtol=1e-3)
def test_quickshift_medoid_estimate():
i = img.copy()
maps, gaps, estimate = quickshift(i,
kernel_size=2,
max_dist=10,
medoid=True)
assert estimate.shape == (512, 512)
assert_allclose(estimate[0:5, 0], [8.699, 11.0754, 12.350, 12.322, 11.190],
rtol=1e-3)
def test_quickshift_rgb():
result = quickshift(img_rgb,
kernel_size=4,
max_dist=24)
assert result[0].dtype == np.float64
assert result[1].dtype == np.float64
assert result[2].dtype == np.float64
assert result[0].shape == img_rgb.shape[:2]
assert result[1].shape == img_rgb.shape[:2]
assert result[2].shape == img_rgb.shape[:2]
def test_quickshift_medoid_gaps():
i = img.copy()
maps, gaps, estimate = quickshift(i,
kernel_size=2,
max_dist=10,
medoid=True)
assert gaps.shape == (512, 512)
assert_allclose(
gaps[0:5,
0], [228071.506, 290406.801, 323886.572, 323339.597, 293392.239],
rtol=1e-3)
def test_flatmap():
maps = quickshift(img, 2, 10)[0]
labels, clusters = flatmap(maps)
assert maps.dtype == np.float64
assert maps.dtype == img.dtype
assert maps.shape == img.shape
assert labels.dtype == maps.dtype
assert labels.shape == maps.shape
assert clusters.dtype == maps.dtype
assert clusters.size == maps.size
def test_quickshift(kernel_size, max_dist, medoid):
result = quickshift(img,
kernel_size=kernel_size,
max_dist=max_dist,
medoid=medoid)
assert result[0].dtype == np.float64
assert result[1].dtype == np.float64
assert result[0].shape == img.shape
assert result[1].shape == img.shape
if max_dist is not None:
assert result[2].dtype == np.float64
assert result[2].shape == img.shape
def test_quickshift_quick_gaps():
i = img.copy()
maps, gaps, estimate = quickshift(i, kernel_size=2, max_dist=10)
assert gaps.shape == (512, 512)
assert_allclose(gaps[0:6, 3], [1., 1., 1., 1.4142, 1., 2.2360], rtol=1e-3)
def test_quickshift_quick_maps():
i = img.copy()
maps, gaps, estimate = quickshift(i, kernel_size=2, max_dist=10)
assert maps.shape == (512, 512)
assert_allclose(maps[0:5, 0], [2., 514., 1026., 1025., 1537.], rtol=1e-3)
import numpy as np
from numpy.testing import assert_allclose
from cyvlfeat.quickshift.quickshift import quickshift
from cyvlfeat.quickshift.flatmap import flatmap
from cyvlfeat.test_util import lena
img = lena().astype(np.float32)
result = quickshift(img, 2, 10)
maps = result[0]
def test_flatmap_labels():
labels, clusters = flatmap(maps)
assert labels.shape == (512, 512)
assert_allclose(labels[0, 0:6],
[1027., 1027., 1027., 1027., 1027., 12303.],
rtol=1e-3)
def test_flatmap_clusters():
labels, clusters = flatmap(maps)
assert clusters.shape == (262144, )
assert_allclose(clusters[0:6], [1., 1., 1., 1., 1., 56.], rtol=1e-3)
def test_quickshift_medoid_maps():
i = img.copy()
maps, gaps, estimate = quickshift(i,
kernel_size=2,
max_dist=10,