def test_threshold2(img):
img.threshold(thresholds=[80, 210], values=[0, 127, 255])
expected = Image.fromArray(
np.array([[127, 255, 0],
[127, 0, 255],
[255, 127, 0]]))
assert img == expected
def test_categorize2(img):
expected = Image.fromArray(
np.array([[255, 255, 100],
[100, 0, 255],
[255, 100, 0]]))
img.categorize(3, initial_clusters=[2, 99, 254])
assert img == expected
def test_threshold1(img):
img.threshold(thresholds=[127], values=[0, 255])
expected = Image.fromArray(
np.array([[255, 255, 0],
[0, 0, 255],
[255, 0, 0]]))
assert img == expected
def test_from_list():
data1 = np.array([[0, 255],
[255, 100]])
data2 = np.array([[100, 255],
[255, 0]])
input_data = [data1, data2]
expected = np.array(input_data)
img = Image.fromArray(input_data).asArray()
assert np.alltrue(img == expected)
def test_threshold():
image = Image.fromArray(np.array([[0.3, 3.0], [1.2, 2.4], [2.1, 1.1]]))
categorical_ref = np.array([[0, 2], [1, 2], [2, 1]])
thresholds = np.array([1.0, 2.0])
image.threshold(thresholds)
labels = mpstool.img.labelize(image)
assert np.alltrue(labels == categorical_ref)
def test_categorize1(img):
expected = Image.fromArray(
np.array([[255, 255, 100],
[100, 100, 255],
[255, 100, 100]]))
print(img._data)
img.categorize(2)
print(img._data)
assert img == expected
def generate(epoch, generator, N, args, device, exportCuts=False):
os.makedirs("output/epoch", exist_ok=True)
for i in range(N):
output = generator.generate(1, device).cpu().detach()
if exportCuts:
cuts = extract_3cuts(output)[0].permute(1, 2, 0)
cuts = (255 * cuts.numpy()).astype(np.uint8)
for ind, dim in enumerate(["x", "y", "z"]):
cut_dim = PILImage.fromarray(cuts[..., ind])
cut_dim.save("output/epoch/{}_{}_{}.png".format(
args.name, i, dim))
output = output.numpy()
output = (output * 255).astype(np.uint8)
output = np.squeeze(output)
output = Image.fromArray(output)
output.exportAsVox("output/epoch/{}_{}.vox".format(args.name, i))
def img():
data = np.array([[200, 255, 60],
[100, 10, 255],
[250, 100, 0]])
return Image.fromArray(data)
def image():
return Image.fromArray(array())
def cube_image(cube):
return Image.fromArray(cube)
min_dict[k] = min_dict[k].clip(0, 1)
max_dict[k] = mean_dict[k] + std_dict[k]
max_dict[k] = max_dict[k].clip(0, 1)
# Read connectivity for the realizations
cX = []
cY = []
cZ = []
cX_mean = {}
cY_mean = {}
cZ_mean = {}
for n in tqdm(range(args.n_samples)):
image = model.generate().to("cpu").detach().numpy() * 255
image = np.squeeze(image).astype(np.uint8)
image = Image.fromArray(image)
image.threshold(thresholds=[254], values=[0, 1])
connZ = stat_fun(image, axis=2)
cZ.append(connZ)
cZ_mean = {k: cZ_mean.get(k, 0) + connZ.get(k, 0) for k in set(connZ)}
connX = stat_fun(image, axis=1)
cX.append(connX)
cX_mean = {k: cX_mean.get(k, 0) + connX.get(k, 0) for k in set(connX)}
connY = stat_fun(image, axis=0)
cY.append(connY)
cY_mean = {k: cY_mean.get(k, 0) + connY.get(k, 0) for k in set(connY)}
cX_mean = {k: cX_mean.get(k, 0) / args.n_samples for k in cX_mean.keys()}