import re
import os
import sys
import random
N = 7
scale = 'color'
def get_cluster_file(cluster):
return 'clusters/%s/%s.jpg' % (N,cluster)
def average(cluster,files):
cmd = 'convert %s -average clusters/%s/%s.jpg' % (' '.join(files),N,cluster)
os.system(cmd)
def dist(file1,file2):
cmd = "compare -metric mae %s %s /dev/null >/tmp/bk_face_dist" % (file1,file2)
os.system(cmd)
f = file('/tmp/bk_face_dist')
s = f.read().strip()
f.close()
s = eval(s.split()[0])
return s
os.system("ls scaled_faces/norm_%s/*.jpg >/tmp/bk_face_dir" % scale)
f = file('/tmp/bk_face_dir')
files = [line.strip() for line in f.readlines()]
f.close()
def init():
global clusters
clusters = [[]]*N
for f in files:
cluster = int(random.random() * N)
clusters[cluster] = clusters[cluster]+[f]
for i in range(N):
average(i,clusters[i])
def iterate():
global clusters
clusters = [[]]*N
for f in files:
closest_i = 0
closest = dist(f,get_cluster_file(0))
for i in range(1,N):
d = dist(f,get_cluster_file(i))
if d < closest:
closest_i = i
closest = d
clusters[closest_i] = clusters[closest_i] + [f]
for i in range(N):
average(i,clusters[i])
init()
for i in range(20):
print i
iterate()
f = file('clusters/%s/clusters.html' % N,'w')
f.write('\n')
for i in range(N):
f.write('%d - 
\n
\n' % (i,i))
for photo in clusters[i]:
match = re.compile("scaled_faces/norm_%s/(.*).jpg" % scale).search(photo)
f.write('
\n' % match.group(1))
f.write('
\n')
f.write('')
f.close()