def label_image(img, cluster_centroids): ''' img: RGB image as 3D list of size [h,w,3] Return a 2D matrix where each point stores the nearest cluster centroid of corresponding pixel in the image [Empty 2D list] -- labels = [[0 for _ in range(width)] for _ in range(height)] ''' cluster_labels = [] # todo: task6 # Iterate over the image pixels and replace each pixel value with cluster number corresponding to its nearest cluster centroid for i in img: temp = [] for j in i: dis = distance(j, cluster_centroids[0]) index = 0 for in_, k in enumerate(cluster_centroids[1:]): t = distance(j, k) if t<dis: dis = t index = in_+1 temp.append(index) cluster_labels.append(temp) return cluster_labels
def label_image(img, cluster_centroids):
'''
img: RGB image as 3D list of size [h,w,3]
Return a 2D matrix where each point stores the nearest cluster centroid of corresponding pixel in the image
[Empty 2D list] -- labels = [[0 for _ in range(width)] for _ in range(height)]
'''
height = len(img)
if len(img) != 0:
width = len(img[0])
cluster_labels = [[0 for _ in range(width)] for _ in range(height)]
for i in range(height):
for l in range(width):
m = 0
d = distance(img[i][l], cluster_centroids[0])
for j in range(len(cluster_centroids)):
d1 = distance(img[i][l], cluster_centroids[j])
if d1 < d:
m = j
d = d1
cluster_labels[i][l] = m
# todo: task6
# Iterate over the image pixels and replace each pixel value with cluster number corresponding to its nearest cluster centroid
return cluster_labels
def label_image(img, cluster_centroids): ''' img: RGB image as 3D list of size [h,w,3] Return a 2D matrix where each point stores the nearest cluster centroid of corresponding pixel in the image [Empty 2D list] -- labels = [[0 for _ in range(width)] for _ in range(height)] ''' cluster_labels = None # todo: task6 # Iterate over the image pixels and replace each pixel value with cluster number corresponding to its nearest cluster centroid cluster_labels = [] for i in range(len(img)): lab = [] for k in range(len(img[0])): min_label = 0 min_dist = distance(img[i][k], cluster_centroids[0]) # tuple(img[i][k]) for j in range(1, len(cluster_centroids)): if(distance(img[i][k], cluster_centroids[j]) < min_dist): min_label = j min_dist = distance(img[i][k], cluster_centroids[j]) lab.append(min_label) cluster_labels.append(lab) return cluster_labels
def label_image(img, cluster_centroids):
'''
img: RGB image as 3D list of size [h,w,3]
Return a 2D matrix where each point stores the nearest cluster centroid of corresponding pixel in the image
[Empty 2D list] -- labels = [[0 for _ in range(width)] for _ in range(height)]
'''
labels = [[0 for _ in range(len(img[0]))] for _ in range(len(img))]
for i in range(len(img)):
for j in range(len(img[0])):
minm = distance(tuple(img[i][j]), cluster_centroids[0])
center = 0
for p in range(1, len(cluster_centroids)):
dist = distance(tuple(img[i][j]), cluster_centroids[p])
if dist < minm:
center = p
minm = dist
labels[i][j] = center
# todo: task6
# Iterate over the image pixels and replace each pixel value with cluster number corresponding to its nearest cluster centroid
return labels
def label_image(img, cluster_centroids):
'''
img: RGB image as 3D list of size [h,w,3]
Return a 2D matrix where each point stores the nearest cluster centroid of corresponding pixel in the image
[Empty 2D list] -- labels = [[0 for _ in range(width)] for _ in range(height)]
'''
cluster_labels = None
# todo: task6
# Iterate over the image pixels and replace each pixel value with cluster number corresponding to its nearest cluster centroid
width = len(img[0])
height = len(img)
data = preprocess_image(img)
cluster_labels = [[0 for _ in range(width)] for _ in range(height)]
i = 0 # height counter
j = 0 # width counter
for pixel in data:
dist = [distance(pixel, centroid) for centroid in cluster_centroids]
min_dist = min(dist)
if (j == width):
i += 1
j = 0
cluster_labels[i][j] = dist.index(min_dist)
j += 1
return cluster_labels
def label_image(img, cluster_centroids):
'''
img: RGB image as 3D list of size [h,w,3]
Return a 2D matrix where each point stores the nearest cluster centroid of corresponding pixel in the image
[Empty 2D list] -- labels = [[0 for _ in range(width)] for _ in range(height)]
'''
h = len(img)
w = len(img[0])
cluster_labels = [[0 for _ in range(w)] for _ in range(h)]
for i in range(h):
for j in range(w):
p = tuple(img[i][j])
mind = float("inf")
mini = 0
for k in range(len(cluster_centroids)):
dist = distance(p, tuple(cluster_centroids[k]))
if (dist < mind):
mind = dist
mini = k
cluster_labels[i][j] = mini
# todo: task6
# Iterate over the image pixels and replace each pixel value with cluster number corresponding to its nearest cluster centroid
return cluster_labels
def label_image(img, cluster_centroids):
'''
img: RGB image as 3D list of size [h,w,3]
Return a 2D matrix where each point stores the nearest cluster centroid of corresponding pixel in the image
[Empty 2D list] -- labels = [[0 for _ in range(width)] for _ in range(height)]
'''
height = len(img)
width = len(img[0])
# todo: task6
# Iterate over the image pixels and replace each pixel value with cluster number corresponding to its nearest cluster centroid
cluster_labels = [[0 for _ in range(width)] for _ in range(height)]
for i in range(height):
for j in range(width):
d_min = -1
d_index = -1
for k in range(len(cluster_centroids)):
d_cluster = distance(cluster_centroids[k], tuple(img[i][j]))
if d_min == -1:
d_min = d_cluster
d_index = k
elif d_cluster < d_min:
d_min = d_cluster
d_index = k
cluster_labels[i][j] = d_index
return cluster_labels
def label_image(img, cluster_centroids): ''' img: RGB image as 3D list of size [h,w,3] Return a 2D matrix where each point stores the nearest cluster centroid of corresponding pixel in the image [Empty 2D list] -- labels = [[0 for _ in range(width)] for _ in range(height)] ''' height = len(img) width = len(img[0]) cluster_labels = [[0 for _ in range(width)] for _ in range(height)] for i in range(height): for j in range(width): darr = [distance(img[i][j], x) for x in cluster_centroids] ind = darr.index(min(darr)) cluster_labels[i][j] = ind # todo: task6 # Iterate over the image pixels and replace each pixel value with cluster number corresponding to its nearest cluster centroid return cluster_labels
def label_image(img, cluster_centroids):
'''
img: RGB image as 3D list of size [h,w,3]
Return a 2D matrix where each point stores the nearest cluster centroid of corresponding pixel in the image
[Empty 2D list] -- labels = [[0 for _ in range(width)] for _ in range(height)]
'''
cluster_labels = None
# todo: task6
# Iterate over the image pixels and replace each pixel value with cluster number corresponding to its nearest cluster centroid
cluster_labels = img
for ind1 in range(len(img)):
for ind2 in range(len(img[ind1])):
dist = [distance(img[ind1][ind2], z) for z in cluster_centroids]
cluster_labels[ind1][ind2] = dist.index(min(dist))
return cluster_labels
def label_image(img, cluster_centroids): ''' img: RGB image as 3D list of size [h,w,3] Return a 2D matrix where each point stores the nearest cluster centroid of corresponding pixel in the image [Empty 2D list] -- labels = [[0 for _ in range(width)] for _ in range(height)] ''' cluster_labels = None # todo: task6 # Iterate over the image pixels and replace each pixel value with cluster number corresponding to its nearest cluster centroid cluster_labels = [ [0 for _ in range(len(img[0]))] for _ in range(len(img))] for i in range(len(img)): for j in range(len(img[0])): datapoint = tuple(img[i][j]) # make a distance list wrt all the cluster points distance_cluster = [ distance(datapoint , k) for k in cluster_centroids] # get the minimal cluster index cluster_labels[i][j] = distance_cluster.index(min(distance_cluster)) return cluster_labels
def label_image(img, cluster_centroids):
'''
img: RGB image as 3D list of size [h,w,3]
Return a 2D matrix where each point stores the nearest cluster centroid of corresponding pixel in the image
[Empty 2D list] -- labels = [[0 for _ in range(width)] for _ in range(height)]
'''
cluster_labels = [[0 for i in range(len(img[j]))] for j in range(len(img))]
# todo: task6
# Iterate over the image pixels and replace each pixel value with cluster number corresponding to its nearest cluster centroid
for i in range(len(img)):
for j in range(len(img[i])):
min_dist = float("inf")
for k in range(len(cluster_centroids)):
dist = distance(img[i][j], cluster_centroids[k])
if dist < min_dist:
min_dist = dist
cluster_labels[i][j] = k
return cluster_labels
def label_image(img, cluster_centroids): ''' img: RGB image as 3D list of size [h,w,3] Return a 2D matrix where each point stores the nearest cluster centroid of corresponding pixel in the image [Empty 2D list] -- labels = [[0 for _ in range(width)] for _ in range(height)] ''' cluster_labels = None height = len(img) width = len(img[0]) cluster_labels = [[0 for _ in range(width)] for _ in range(height)] for i in range(height): for j in range(width): # print(img[i][j]) distancelist = map(lambda x: distance(img[i][j],x),cluster_centroids) # print(distancelist.index(min(distancelist))) cluster_labels[i][j] = distancelist.index(min(distancelist)) # todo: task6 # Iterate over the image pixels and replace each pixel value with cluster number corresponding to its nearest cluster centroid return cluster_labels
def label_image(img, cluster_centroids):
'''
img: RGB image as 3D list of size [h,w,3]
Return a 2D matrix where each point stores the nearest cluster centroid of corresponding pixel in the image
[Empty 2D list] -- labels = [[0 for _ in range(width)] for _ in range(height)]
'''
cluster_labels = []
height = len(img)
width = len(img[0])
for i in range(height):
row = []
for j in range(width):
temp = [
distance(centroid, img[i][j]) for centroid in cluster_centroids
]
index = temp.index(min(temp))
row.append(index)
cluster_labels.append(row)
# todo: task6
# Iterate over the image pixels and replace each pixel value with cluster number corresponding to its nearest cluster centroid
return cluster_labels
def test_for_cluster():
for_node = [(i, 'I') for i in range(1, 4)]
for_node += [(i, 'O') for i in range(4, 6)]
for_node += [(i, 'H2') for i in range(6, 8)]
node_ctr = 8
indim = 3
outdim = 2
innov_num = 11
dob = 0
node_lis = [gene.Node(x, y) for x, y in for_node]
for_conn = [(1, (1, 4), 0.4, False), (2, (1, 5), 0.25, True),
(3, (2, 4), 0.25, True), (4, (2, 5), 0.5, True),
(5, (3, 4), 0.7, True), (6, (3, 5), 0.6, False),
(7, (1, 6), 0.5, True), (8, (6, 4), 0.4, True),
(9, (3, 7), 0.3, True), (10, (7, 5), 0.6, True)]
conn_lis = [
gene.Conn(x, (node_lis[tup[0] - 1], node_lis[tup[1] - 1]), w, status)
for x, tup, w, status in for_conn
]
for_bias = [(4, 0.2), (5, 0.1)]
bias_conn_lis = [gene.BiasConn(node_lis[x - 1], y) for x, y in for_bias]
newchromo1 = Chromosome(indim, outdim)
newchromo1.__setattr__('conn_arr', conn_lis)
newchromo1.__setattr__('bias_conn_arr', bias_conn_lis)
newchromo1.__setattr__('node_arr', node_lis)
newchromo1.__setattr__('dob', dob)
newchromo1.set_node_ctr(node_ctr)
for_node = [(i, 'I') for i in range(1, 4)]
for_node += [(i, 'O') for i in range(4, 6)]
st = '2212211'
for_node += [(i + 6, 'H' + st[i]) for i in range(len(st))]
node_ctr = 13
innov_num = 25
dob = 0
indim = 8
outdim = 2
node_lis = [gene.Node(x, y) for x, y in for_node]
for_conn = [
(1, (1, 4), 0.3, True),
(2, (1, 5), 0.25, False),
(3, (2, 4), 0.25, False),
(4, (2, 5), 0.5, False),
(5, (3, 4), 0.7, False),
(6, (3, 5), 0.5, True),
(7, (1, 6), 0.2, True),
(8, (6, 4), 0.1, True),
(9, (2, 7), 0.1, True),
(10, (7, 4), 0.15, True),
(11, (1, 8), 0.5, True),
(12, (8, 6), 0.7, True),
(13, (1, 9), 0.3, False),
(14, (9, 5), 1.0, True),
(15, (3, 10), 0.33, True),
(16, (10, 5), 0.77, True),
(17, (1, 11), 0.25, True),
(18, (11, 9), 0.15, True),
(19, (2, 12), 0.6, True),
(20, (12, 7), 0.4, True),
(21, (3, 12), 0.8, True),
(22, (2, 9), 0.9, True),
(23, (12, 4), 0.75, True),
(24, (11, 5), 0.25, True),
]
conn_lis = [
gene.Conn(x, (node_lis[tup[0] - 1], node_lis[tup[1] - 1]), w, status)
for x, tup, w, status in for_conn
]
for_bias = [(4, 0.2), (5, 0.1)]
bias_conn_lis = [gene.BiasConn(node_lis[x - 1], y) for x, y in for_bias]
newchromo2 = Chromosome(indim, outdim)
newchromo2.__setattr__('conn_arr', conn_lis)
newchromo2.__setattr__('bias_conn_arr', bias_conn_lis)
newchromo2.__setattr__('node_arr', node_lis)
newchromo2.__setattr__('dob', dob)
newchromo2.set_node_ctr(node_ctr)
# newchromo.pp()
for_node = [(i, 'I') for i in range(1, 4)]
for_node += [(i, 'O') for i in range(4, 6)]
st = '221221'
for_node += [(i + 6, 'H' + st[i]) for i in range(len(st))]
node_ctr = 12
dob = 0
indim = 3
outdim = 2
node_lis = [gene.Node(x, y) for x, y in for_node]
for_conn = [(1, (1, 4), 0.3, True), (2, (1, 5), 0.25, False),
(3, (2, 4), 0.25, False), (4, (2, 5), 0.5, False),
(5, (3, 4), 0.7, False), (6, (3, 5), 0.5, True),
(7, (1, 6), 0.2, True), (8, (6, 4), 0.1, True),
(9, (2, 7), 0.1, True), (10, (7, 4), 0.15, True),
(11, (1, 8), 0.5, True), (12, (8, 6), 0.7, True),
(13, (1, 9), 0.3, False), (14, (9, 5), 1.0, True),
(15, (3, 10), 0.33, True), (16, (10, 5), 0.77, True),
(19, (2, 11), 0.6, True), (20, (11, 7), 0.4, True),
(22, (3, 11), 0.75, True)]
conn_lis = [
gene.Conn(x, (node_lis[tup[0] - 1], node_lis[tup[1] - 1]), w, status)
for x, tup, w, status in for_conn
]
for_bias = [(4, 0.2), (5, 0.1)]
bias_conn_lis = [gene.BiasConn(node_lis[x - 1], y) for x, y in for_bias]
newchromo3 = Chromosome(indim, outdim)
newchromo3.__setattr__('conn_arr', conn_lis)
newchromo3.__setattr__('bias_conn_arr', bias_conn_lis)
newchromo3.__setattr__('node_arr', node_lis)
newchromo3.__setattr__('dob', dob)
newchromo3.set_node_ctr(node_ctr)
for_node = [(i, 'I') for i in range(1, 4)]
for_node += [(i, 'O') for i in range(4, 6)]
st = '221221'
for_node += [(i + 6, 'H' + st[i]) for i in range(len(st))]
node_ctr = 12
innov_num = 21
dob = 0
indim = 3
outdim = 2
node_lis = [gene.Node(x, y) for x, y in for_node]
for_conn = [(1, (1, 4), 0.3, True), (2, (1, 5), 0.25, False),
(3, (2, 4), 0.25, False), (4, (2, 5), 0.5, False),
(5, (3, 4), 0.7, False), (6, (3, 5), 0.5, True),
(7, (1, 6), 0.2, True), (8, (6, 4), 0.1, True),
(9, (2, 7), 0.1, True), (10, (7, 4), 0.15, True),
(11, (1, 8), 0.5, True), (12, (8, 6), 0.7, True),
(13, (1, 9), 0.3, False), (14, (9, 5), 1.0, True),
(15, (3, 10), 0.33, True), (16, (10, 5), 0.77, True),
(17, (1, 11), 0.25, True), (18, (11, 9), 0.15, True),
(21, (2, 9), 0.65, True)]
conn_lis = [
gene.Conn(x, (node_lis[tup[0] - 1], node_lis[tup[1] - 1]), w, status)
for x, tup, w, status in for_conn
]
for_bias = [(4, 0.2), (5, 0.1)]
bias_conn_lis = [gene.BiasConn(node_lis[x - 1], y) for x, y in for_bias]
newchromo4 = Chromosome(indim, outdim)
newchromo4.__setattr__('conn_arr', conn_lis)
newchromo4.__setattr__('bias_conn_arr', bias_conn_lis)
newchromo4.__setattr__('node_arr', node_lis)
newchromo4.__setattr__('dob', dob)
newchromo4.set_node_ctr(node_ctr)
for_node = [(i, 'I') for i in range(1, 4)]
for_node += [(i, 'O') for i in range(4, 6)]
st = '2212211'
for_node += [(i + 6, 'H' + st[i]) for i in range(len(st))]
node_ctr = 13
innov_num = 25
dob = 0
indim = 3
outdim = 2
node_lis = [gene.Node(x, y) for x, y in for_node]
for_conn = [
(1, (1, 4), 0.3, False),
(2, (1, 5), 0.25, False),
(3, (2, 4), 0.25, False),
(4, (2, 5), 0.5, False),
(5, (3, 4), 0.7, False),
(6, (3, 5), 0.5, True),
(7, (1, 6), 0.2, True),
(8, (6, 4), 0.1, True),
(9, (2, 7), 0.1, True),
(10, (7, 4), 0.15, True),
(11, (1, 8), 0.5, True),
(12, (8, 6), 0.7, True),
(13, (1, 9), 0.3, False),
(14, (9, 5), 1.0, True),
(15, (3, 10), 0.33, True),
(16, (10, 5), 0.77, True),
(17, (1, 11), 0.25, True),
(21, (3, 12), 0.8, True),
(22, (2, 9), 0.9, True),
(23, (12, 4), 0.75, True),
(24, (11, 5), 0.25, True),
]
conn_lis = [
gene.Conn(x, (node_lis[tup[0] - 1], node_lis[tup[1] - 1]), w, status)
for x, tup, w, status in for_conn
]
for_bias = [(4, 0.2), (5, 0.1)]
bias_conn_lis = [gene.BiasConn(node_lis[x - 1], y) for x, y in for_bias]
newchromo5 = Chromosome(indim, outdim)
newchromo5.__setattr__('conn_arr', conn_lis)
newchromo5.__setattr__('bias_conn_arr', bias_conn_lis)
newchromo5.__setattr__('node_arr', node_lis)
newchromo5.__setattr__('dob', dob)
newchromo5.set_node_ctr(node_ctr)
for_node = [(i, 'I') for i in range(1, 4)]
for_node += [(i, 'O') for i in range(4, 6)]
st = '22122'
for_node += [(i + 6, 'H' + st[i]) for i in range(len(st))]
node_ctr = 11
innov_num = 17
dob = 0
indim = 3
outdim = 2
node_lis = [gene.Node(x, y) for x, y in for_node]
for_conn = [(1, (1, 4), 0.8, True), (2, (1, 5), 0.25, False),
(3, (2, 4), 0.25, False), (4, (2, 5), 0.5, False),
(5, (3, 4), 0.7, False), (6, (3, 5), 0.5, True),
(7, (1, 6), 0.2, True), (8, (6, 4), 0.1, True),
(9, (2, 7), 0.1, True), (10, (7, 4), 0.15, True),
(11, (1, 8), 0.5, True), (12, (8, 6), 0.7, True),
(13, (1, 9), 0.3, False), (14, (9, 5), 1.0, True),
(15, (3, 10), 0.33, True), (16, (10, 5), 0.77, True)]
conn_lis = [
gene.Conn(x, (node_lis[tup[0] - 1], node_lis[tup[1] - 1]), w, status)
for x, tup, w, status in for_conn
]
for_bias = [(4, 0.2), (5, 0.1)]
bias_conn_lis = [gene.BiasConn(node_lis[x - 1], y) for x, y in for_bias]
newchromo6 = Chromosome(indim, outdim)
newchromo6.__setattr__('conn_arr', conn_lis)
newchromo6.__setattr__('bias_conn_arr', bias_conn_lis)
newchromo6.__setattr__('node_arr', node_lis)
newchromo6.__setattr__('dob', dob)
newchromo6.set_node_ctr(node_ctr)
for_node = [(i, 'I') for i in range(1, 4)]
for_node += [(i, 'O') for i in range(4, 6)]
for_node += [(i, 'H2') for i in range(6, 8)]
node_ctr = 8
indim = 3
outdim = 2
innov_num = 11
dob = 0
node_lis = [gene.Node(x, y) for x, y in for_node]
for_conn = [(1, (1, 4), 0.4, False), (2, (1, 5), 0.25, True),
(3, (2, 4), 0.25, True), (4, (2, 5), 0.5, True),
(5, (3, 4), 0.7, True), (6, (3, 5), 0.6, False),
(7, (1, 6), 0.5, True), (8, (6, 4), 0.4, True),
(9, (3, 7), 0.3, True)]
conn_lis = [
gene.Conn(x, (node_lis[tup[0] - 1], node_lis[tup[1] - 1]), w, status)
for x, tup, w, status in for_conn
]
for_bias = [(4, 0.2), (5, 0.1)]
bias_conn_lis = [gene.BiasConn(node_lis[x - 1], y) for x, y in for_bias]
newchromo7 = Chromosome(indim, outdim)
newchromo7.__setattr__('conn_arr', conn_lis)
newchromo7.__setattr__('bias_conn_arr', bias_conn_lis)
newchromo7.__setattr__('node_arr', node_lis)
newchromo7.__setattr__('dob', dob)
newchromo7.set_node_ctr(node_ctr)
chromo_list = [
newchromo1, newchromo2, newchromo3, newchromo4, newchromo5, newchromo6,
newchromo7
]
#print([item.pp() for item in chromo_list])
print(cluster.distance(newchromo3, newchromo2))
cluster.give_cluster_head(chromo_list, 2)
inputarr = np.array([[0, 2, 1], [0.8, 1, 2]])
def get_closest_centroid_idx(point):
centroid_distances = map(lambda c: distance(point, c),
cluster_centroids)
return centroid_distances.index(min(centroid_distances))
