def calculate_cluster(lena, lena_mat, quantile):
bandwidth = estimate_bandwidth(lena_mat, quantile=quantile, n_samples=500)
ms = MeanShift(bandwidth=bandwidth, bin_seeding=True)
ms.fit(lena_mat)
labels = ms.labels_
cluster_centers = ms.cluster_centers_
labels_unique = np.unique(labels)
n_clusters_ = len(labels_unique)
lena_clustered = lena.copy()
lena_clustered_value = lena.copy()
lena_mat_clustered = lena_mat.copy()
lena_mat_clustered_value = lena_mat.copy()
for point, pointb, value in zip(lena_mat_clustered, lena_mat_clustered_value, labels):
point[2] = value
pointb[2] = cluster_centers[value, 2]
lena_clustered[point[0], point[1]] = value
lena_clustered_value[point[0], point[1]] = cluster_centers[value, 2]
image = {"image": lena_clustered_value,
"quantile": quantile,
"clusters": n_clusters_}
return image
def calculate_cluster(camera, camera_mat, quantile):
bandwidth = estimate_bandwidth(camera_mat,
quantile=quantile,
n_samples=500)
ms = MeanShift(bandwidth=bandwidth, bin_seeding=True)
ms.fit(camera_mat)
labels = ms.labels_
cluster_centers = ms.cluster_centers_
labels_unique = np.unique(labels)
n_clusters_ = len(labels_unique)
camera_clustered = camera.copy()
camera_clustered_value = camera.copy()
camera_mat_clustered = camera_mat.copy()
camera_mat_clustered_value = camera_mat.copy()
for point, pointb, value in zip(camera_mat_clustered,
camera_mat_clustered_value, labels):
point[2] = value
pointb[2] = cluster_centers[value, 2]
camera_clustered[point[0], point[1]] = value
camera_clustered_value[point[0], point[1]] = cluster_centers[value, 2]
image = {
"image": camera_clustered_value,
"quantile": quantile,
"clusters": n_clusters_
}
return image
def meanshift(desc, quantile, hs=16, hr=16, copy=True):
"""
Do nothing for now...
"""
if copy:
desc = desc.copy()
desc[:, :2] /= hs
desc[:, 2:] /= hr
bandwidth = estimate_bandwidth(desc, quantile=quantile, n_samples=500)
ms = MeanShift(bandwidth=bandwidth, bin_seeding=True)
ms.fit(desc)
ms.cluster_centers_[:, :2] *= hs
ms.cluster_centers_[:, 2:] *= hr
return ms
from collections import defaultdict
import numpy as np
from sklearn.datasets import make_blobs
from sklearn.cluster.mean_shift_ import estimate_bandwidth
from ipywidgets import interactive_output, IntSlider, Play, jslink
import matplotlib.pyplot as plt
from matplotlib.figure import figaspect
from matplotlib.patches import Circle
min_bin_freq = 1
X, _ = make_blobs(centers=3,
cluster_std=1,
center_box=(-7, 7),
shuffle=False,
random_state=1234)
bandwidth = estimate_bandwidth(X, n_jobs=-1)
bin_sizes = defaultdict(int)
for point in X:
binned_point = np.round(point / bandwidth)
bin_sizes[tuple(binned_point)] += 1
lowers = np.round(X.min(axis=0) / bandwidth)
uppers = np.round(X.max(axis=0) / bandwidth)
xbounds = np.arange(lowers[0] + 0.5, uppers[0] + 1.5) * bandwidth
ybounds = np.arange(lowers[1] + 0.5, uppers[1] + 1.5) * bandwidth
bin_seeds = np.array(
[point for point, freq in bin_sizes.items() if freq >= min_bin_freq],
dtype=np.float32) * bandwidth
