def load_external_hit(url="http://65.215.1.20/faces/simp_sim3.py",
assignments=1,
input="1",
title="A nice HIT",
description="Please help with our AI task",
reward="0.15",
duration=900,
autoapproval=259200,
approvalrate=95,
height=500,
keywords="image, similarity",
numapproved=48,
comment=None):
ell = locals()
label = get_next_label()
input2 = log_root + "/" + label + ".input"
question2 = log_root + "/" + label + ".question"
properties2 = log_root + "/" + label + ".properties"
itmpl = tools.my_read(input_template)
qtmpl = tools.my_read(question_template)
ptmpl = tools.my_read(properties_template)
st = string.Template(itmpl).substitute(ell)
tools.my_write(input2, input_shortcutifier(st))
st = string.Template(qtmpl).substitute(ell)
tools.my_write(question2, st)
st = string.Template(ptmpl).substitute(ell)
tools.my_write(properties2, st)
return load_hit_helper(label, input2, question2, properties2, comment)
def load_external_hit(url="http://65.215.1.20/faces/simp_sim3.py",assignments=1,input="1",title="A nice HIT",description="Please help with our AI task",reward="0.15",duration=900,autoapproval=259200,approvalrate=95,height=500,keywords="image, similarity",numapproved=48,comment=None):
ell = locals()
label = get_next_label()
input2 = log_root+"/"+label+".input"
question2 = log_root+"/"+label+".question"
properties2 = log_root+"/"+label+".properties"
itmpl=tools.my_read(input_template)
qtmpl=tools.my_read(question_template)
ptmpl=tools.my_read(properties_template)
st = string.Template(itmpl).substitute(ell)
tools.my_write(input2,input_shortcutifier(st))
st = string.Template(qtmpl).substitute(ell)
tools.my_write(question2,st)
st = string.Template(ptmpl).substitute(ell)
tools.my_write(properties2,st)
return load_hit_helper(label,input2,question2,properties2, comment)
def get_agreement(filename):
s = tools.my_read(filename).strip().splitlines()
trips = {}
for l in s:
(a, b, c, e) = l.split()
a = int(a)
b = int(b)
c = int(c)
tb = min(b, c)
tc = max(b, c)
if b < c:
y = 1
else:
y = 0
if (a, tb, tc) not in trips:
trips[(a, tb, tc)] = [0, 0]
trips[(a, tb, tc)][y] = trips[(a, tb, tc)][y] + 1
v = trips.values()
data = []
for (a, b) in v:
if (a + b > 1):
data.append((a * (a - 1.) + b * (b - 1.)) / (
(a + b) *
(a + b - 1.))) # double count agreement and disagreement pairs
r = scipy.stats.morestats.bayes_mvs(data, 0.95)[0]
print "Mean", r[0], " 95% confidence interval", r[1]
print "This corresponds to a signal of ", 0.5 + np.sqrt(r[0] / 2. - 1 / 4.)
return r[0]
def read_out_files(fnames):
if type(fnames)!=list:
fnames = [fnames]
res = []
for f in fnames:
res += [map(int,i.split()[:3]) for i in tools.my_read(f).strip().splitlines()]
return res
def create_info(dir):
if dir[-1] != "/" and dir[-1] != "\\":
dir += "/"
assert not os.path.exists(dir + "info.csv")
ids = tools.my_read(dir + "ids.txt").strip().splitlines()
n = len(ids)
print n, "ids."
rs = []
atts = [i for i in os.listdir(dir) if "." not in i]
print "Attributes:", atts
for i in range(n):
r = {"id": i, "image": ids[i]}
for att in atts:
if att[:4] != "att_":
the_att = "att_" + att
else:
the_att = att
if os.path.exists(dir + att + "/1/" + ids[i]):
r[the_att] = 1
else:
if os.path.exists(dir + att + "/0/" + ids[i]):
r[the_att] = 0
else:
r[the_att] = -100
rs.append(r)
f = open(dir + "info.csv", "wb")
c = csv.DictWriter(f, sorted(rs[0].keys()))
header = {}
for k in rs[0].keys():
header[k] = k
c.writerow(header)
c.writerows(rs)
f.close()
def get_agreement(filename):
s = tools.my_read(filename).strip().splitlines()
trips = {}
for l in s:
(a,b,c,e) = l.split()
a = int(a)
b = int(b)
c = int(c)
tb = min(b,c)
tc = max(b,c)
if b<c:
y = 1
else:
y = 0
if (a,tb,tc) not in trips:
trips[(a,tb,tc)]=[0,0]
trips[(a,tb,tc)][y]=trips[(a,tb,tc)][y]+1
v = trips.values()
data = []
for (a,b) in v:
if(a+b>1):
data.append( (a*(a-1.)+b*(b-1.))/((a+b)*(a+b-1.))) # double count agreement and disagreement pairs
r =scipy.stats.morestats.bayes_mvs(data, 0.95)[0]
print "Mean",r[0], " 95% confidence interval",r[1]
print "This corresponds to a signal of ", 0.5+np.sqrt(r[0]/2.-1/4.)
return r[0]
def create_info(dir):
if dir[-1]!="/" and dir[-1]!="\\":
dir+="/"
assert not os.path.exists(dir+"info.csv")
ids = tools.my_read(dir+"ids.txt").strip().splitlines()
n = len(ids)
print n,"ids."
rs = []
atts = [i for i in os.listdir(dir) if "." not in i]
print "Attributes:",atts
for i in range(n):
r = { "id": i, "image": ids[i] }
for att in atts:
if att[:4]!="att_":
the_att="att_"+att
else:
the_att = att
if os.path.exists(dir+att+"/1/"+ids[i]):
r[the_att] = 1
else:
if os.path.exists(dir+att+"/0/"+ids[i]):
r[the_att] = 0
else:
r[the_att] = -100
rs.append(r)
f=open(dir+"info.csv","wb")
c = csv.DictWriter(f,sorted(rs[0].keys()))
header = {}
for k in rs[0].keys():
header[k]=k
c.writerow(header)
c.writerows(rs)
f.close()
def out2trips(filename):
lst = []
lines = tools.my_read(filename).splitlines()
for l in lines:
(a,b,c,d) = l.split()
if a!=b and a!=c:
lst.append((int(a),int(b),int(c)))
else:
print "dumb trip"
return lst
def py_to_server(label="test", pycontents=None):
if pycontents == None:
pycontents = tools.my_read(exp_root + "/to_server.py")
label = str(label)
url = "http://65.215.1.20/faces/add_exp.py?"
url += urllib.urlencode([("name", label), ("contents", pycontents)])
f = urllib2.urlopen(url)
r = f.read().strip()
assert r == "<html>SUCCESS</html>"
return "http://65.215.1.20/faces/exps/" + label + ".py"
def out2trips(filename):
lst = []
lines = tools.my_read(filename).splitlines()
for l in lines:
(a, b, c, d) = l.split()
if a != b and a != c:
lst.append((int(a), int(b), int(c)))
else:
print "dumb trip"
return lst
def read_out_files(fnames):
if type(fnames) != list:
fnames = [fnames]
res = []
for f in fnames:
res += [
map(int,
i.split()[:3]) for i in tools.my_read(f).strip().splitlines()
]
return res
def py_to_server(label="test",pycontents=None):
if pycontents == None:
pycontents = tools.my_read(exp_root+"/to_server.py")
label=str(label)
url = "http://65.215.1.20/faces/add_exp.py?"
url += urllib.urlencode([("name",label),("contents",pycontents)])
f=urllib2.urlopen(url)
r = f.read().strip()
assert r=="<html>SUCCESS</html>"
return "http://65.215.1.20/faces/exps/"+label+".py"
def read_model(model_file):
d = tools.my_read(model_file).strip().splitlines()
assert(d[9][-20:].strip()=="port vectors plus 1")
num_vectors = int(d[9].split()[0])-1
thresh = float(d[10].split()[0])
#print num_vectors, thresh
alphas = np.array([[float(t.split()[0]) for t in d[11:]]])
#print len(alphas)
svs = np.array([[float(t.split(":")[1]) for t in r.strip().split()[1:-2]] for r in d[11:]])
w = np.dot(alphas,svs)
return w, thresh
def test_omers_fit():
print "Testing Omer's neckties"
adaptive_trips = read_out_files(glob.glob("c:/sim/turkexps/neckties/small/*.out"))
print len(adaptive_trips),"adaptive trips"
random_trips = read_out_files(glob.glob("c:/sim/turkexps/neckties/small/random/*.out"))
print len(random_trips),"random trips"
control_trips = read_out_files(glob.glob("c:/sim/turkexps/neckties/small/control/*.out"))
print len(control_trips),"control trips"
M = np.array([[float(b) for b in a.split()] for a in tools.my_read("c:/users/adum/simexp/matlab/neckties_fit_5.txt").strip().splitlines()])
S = np.dot(M,M.T)
#grad_proj2(random_trips,control_trips,step_size=5,its=2,Sinit=S)
print "n=",S.shape[0]
show_nn(S)
show_posts(S,random_trips)
doppel_trips(S,random_trips) #also loads posts from file
def test_omers_fit():
print "Testing Omer's neckties"
adaptive_trips = read_out_files(
glob.glob("c:/sim/turkexps/neckties/small/*.out"))
print len(adaptive_trips), "adaptive trips"
random_trips = read_out_files(
glob.glob("c:/sim/turkexps/neckties/small/random/*.out"))
print len(random_trips), "random trips"
control_trips = read_out_files(
glob.glob("c:/sim/turkexps/neckties/small/control/*.out"))
print len(control_trips), "control trips"
M = np.array([[float(b) for b in a.split()] for a in tools.my_read(
"c:/users/adum/simexp/matlab/neckties_fit_5.txt").strip().splitlines()
])
S = np.dot(M, M.T)
#grad_proj2(random_trips,control_trips,step_size=5,its=2,Sinit=S)
print "n=", S.shape[0]
show_nn(S)
show_posts(S, random_trips)
doppel_trips(S, random_trips) #also loads posts from file
def create_data(dir,att,info_file,fit_file,data_file):
x = tools.my_read(fit_file).strip().splitlines()
x = [[float(zz) for zz in z.split()] for z in x]
m = len(x)
d = len(x[0])
f = open(info_file,"r")
csvr = csv.DictReader(f)
y = [int(r[att]) for r in csvr]
f.close()
st = ""
for i in range(m):
if y[i]!=-100:
if y[i]==1:
st+="+1"
else: #translate 0 to -1
st+="-1"
for j in range(d):
st+=" "+str(j+1)+":"+str(x[i][j])
st+=" # "+str(i)
st+="\n"
st = st.strip()
tools.my_write(data_file,st)
return (x,y)
#print "or"
#print "checktrips.py batch_id, where batch_id is the integer id of the batch"
#print "or"
#print "checktrips.py outfileloc"
if len(sys.argv) == 2:
label = sys.argv[1].strip()
else:
label = str(turk.get_last_label())
if "\\" in label or "." in label or "/" in label:
label = label.replace("/", "\\")
count = 0
fname = ""
for filename in glob.glob(turk.log_root + "/*.out_file_loc"):
if tools.my_read(filename).strip().replace("/", "\\") == label:
fname = filename
count += 1
if count > 1:
print "Found more than one .out_file_loc file pointing to", label
print "Exiting..."
exit(0)
if count == 0:
print "Couldn't find any .out_file_loc file pointing to", label
print "Maybe try absolute path"
exit(0)
label = fname[fname.rindex("\\") + 1:-13]
print label
trips = tools.remove_comments(
tools.my_read(turk.log_root + "/" + label + ".trips").strip().split("\n"))
if len(sys.argv)!=4 and len(sys.argv)!=3:
print "Usage: runtrips.py config_file trips_file output_file"
print " or"
print "Usage: runtrips.py config_file trips_file"
print "in which case the output file is trips_file.out"
exit(1)
if len(sys.argv)==3:
(a,b,c) = sys.argv[2].rpartition(".")
if b=="":
sys.argv.append(sys.argv[2]+".out")
else:
sys.argv.append(a+".out")
print "outfile "+sys.argv[-1]
config = tools.remove_comments(tools.my_read(sys.argv[1]).split("\n"))
trips = tools.remove_comments(tools.my_read(sys.argv[2]).strip().split("\n"))
db = config[0]
payment = config[1]
assert float(payment)<0.9 #make sure we don't pay more than 90 cents/hit!
height = config[2]
comment = config[3]
#see if line 4 has an integer
try:
reps = int(config[4])
instructions = string.join(config[5:],"\n")
except:
reps = 1
instructions = string.join(config[4:],"\n")
s+='var datab = "'+db+'";\n'
def list_to_str(lst):
res = "["
for i in lst:
if type(i)==list:
res+=list_to_str(i)
else:
res+=str(i)
res+=","
if res[-1]==",":
return res[:-1] + "]"
return res+"]"
tree = []
inf = tools.my_read(infile).splitlines()
for line in inf:
(a,b) = line.split(":")
tree.append([a.split(),b.split()])
s+='var tree = '+list_to_str(tree)+";\n"
s+='var ids = ['
ids = tools.my_read(dataroot+"/"+db+"/ids.txt").splitlines()
for i in ids:
s+='"'+i+'",'
s=s[:-1]
s+='];\n'
s+= '</script></head>\n'
s+= tools.my_read(htmlfile)
tools.my_write(outfile,s)
def im_filename(j):
global filenames
if filenames == None:
filenames = tools.my_read(fdir + "/ids.txt").strip().splitlines()
return fdir + "\\" + filenames[j]
@author: adum
"""
import tools, turk, csv
from numpy import *
import matplotlib.pyplot as plt
import Tkinter
from PIL import Image, ImageTk
import random
#trip_file = "c:/sim/turkexps/fonts/fonts1.out"
trip_file = "c:/sim/fonts10000.out"
ids = []
id_locs = tools.my_read("c:/matlab/fonts/ids.txt").strip().split("\n")
txt = tools.my_read(trip_file).strip().split("\n")
txt2 = [l.split() for l in txt]
trips = [(int(l[0]), int(l[1]), int(l[2])) for l in txt2]
for t in trips:
for i in t:
if i not in ids:
ids.append(i)
trips2 = [(ids.index(a), ids.index(b), ids.index(c)) for (a, b, c) in trips]
n = len(ids)
print n
# -*- coding: utf-8 -*-
"""
Created on Wed Oct 06 08:08:39 2010
@author: adum
"""
import tools, numpy
import random
coords = tools.my_read("c:/sim/tsp/fonts.txt").strip().splitlines()
M = []
for x in coords:
(a,b,c)=x.split()
M.append((float(a),float(b),float(c)))
print len(M)
def draw_fonts(f,title):
res="<table width=300 border=1>"
res+="<tr><td><u>"+title+"</u></td></tr>"
for i in f:
res+='<tr><td><font face="'+i+'">'+i+'</font></td></tr>\n'
res+="</table>\n"
return res
def font_dist(i,j):
# if i==5:
# return 0
def list_to_str(lst):
res = "["
for i in lst:
if type(i) == list:
res += list_to_str(i)
else:
res += str(i)
res += ","
if res[-1] == ",":
return res[:-1] + "]"
return res + "]"
tree = []
inf = tools.my_read(infile).splitlines()
for line in inf:
(a, b) = line.split(":")
tree.append([a.split(), b.split()])
s += 'var tree = ' + list_to_str(tree) + ";\n"
s += 'var ids = ['
ids = tools.my_read(dataroot + "/" + db + "/ids.txt").splitlines()
for i in ids:
s += '"' + i + '",'
s = s[:-1]
s += '];\n'
s += '</script></head>\n'
s += tools.my_read(htmlfile)
tools.my_write(outfile, s)
def im_filename(j):
global filenames
if filenames==None:
filenames = tools.my_read(fdir+"/ids.txt").strip().splitlines()
return fdir+"\\"+filenames[j]
]
def remove_comments(s):
if type(s) == list:
return [remove_comments(i) for i in s]
if s.find("#") != -1:
return s[:s.index("#")].strip()
return s.strip()
if len(sys.argv) != 4:
print "Usage: runtrips.py config_file trips_file output_file"
exit(1)
config = remove_comments(tools.my_read(sys.argv[1]).split("\n"))
trips = remove_comments(tools.my_read(sys.argv[2]).strip().split("\n"))
db = config[0]
payment = config[1]
assert float(payment) < 0.9 #make sure we don't pay more than 90 cents/hit!
height = config[2]
comment = config[3]
instructions = string.join(config[4:], "\n")
if "%s" not in instructions:
print "Instructions should have %s in them!"
inst_shortcut = turk.shortcut_to_server(instructions)
trips = [i.split() for i in trips]
#print trips
@author: adum
"""
import tools, turk, csv
from numpy import *
import matplotlib.pyplot as plt
import Tkinter
from PIL import Image, ImageTk
import random
#trip_file = "c:/sim/turkexps/fonts/fonts1.out"
trip_file = "c:/sim/fonts10000.out"
ids = []
id_locs = tools.my_read("c:/matlab/fonts/ids.txt").strip().split("\n")
txt = tools.my_read(trip_file).strip().split("\n")
txt2 = [l.split() for l in txt]
trips = [(int(l[0]),int(l[1]),int(l[2])) for l in txt2]
for t in trips:
for i in t:
if i not in ids:
ids.append(i)
trips2 = [ (ids.index(a),ids.index(b),ids.index(c)) for (a,b,c) in trips]
n = len(ids)
print n
#print "or"
#print "checktrips.py batch_id, where batch_id is the integer id of the batch"
if len(sys.argv)==2:
label = sys.argv[1].strip()
else:
label = str(turk.get_last_label())
if "\\" in label or "." in label or "/" in label:
label=label.replace("/","\\")
count = 0
fname = ""
for filename in glob.glob(turk.log_root+"/*.out_file_loc"):
if tools.my_read(filename).strip().replace("/","\\")==label:
fname = filename
count += 1
if count>1:
print "Found more than one .out_file_loc file pointing to",label
print "Exiting..."
exit(0)
if count==0:
print "Couldn't find any .out_file_loc file pointing to",label
print "Maybe try absolute path"
exit(0)
label = fname[fname.rindex("\\")+1:-13]
print label
trips = tools.remove_comments(tools.my_read(turk.log_root+"/"+label+".trips").strip().split("\n"))