Code pour la cross-validation

Author: remi.paucher

Hotmail_2_GenerateDB.py

@@ -23,8 +23,8 @@ def Myexcepthook(type, value, tb):
 GENERATE_VALIDATION_SET = False
 
 FONTS = [
-        ("Fonts/califb.ttf", (180,)),
-         ("Fonts/georgia.ttf", (180,)),
+         ("Fonts/califb.ttf", (180,)),
+         #("Fonts/georgia.ttf", (180,)),
          ("Fonts/sylfaen.ttf", (180,)),
          #("Fonts/BKANT.TTF", (180,)),
          ]
@@ -37,19 +37,18 @@ def Myexcepthook(type, value, tb):
     ##############################################################################
     """
 
-    #GENERATE_CAPITAL_LETTERS = True
-    #GENERATE_DIGITS = True
-    #elem_to_gen = Generate_Element_List(GENERATE_CAPITAL_LETTERS, GENERATE_DIGITS)
-    
-    #elem_to_gen = ['B', 'D', 'E', 'M', 'N', 'T', 'U', '2', '3', '8']
-    elem_to_gen = '3DE2MT'
+    GENERATE_CAPITAL_LETTERS = True
+    GENERATE_DIGITS = True
+    elem_to_gen = Generate_Element_List(GENERATE_CAPITAL_LETTERS, GENERATE_DIGITS)
+    
+    #elem_to_gen = '3DE2MT'
 
     DESTINATION_FOLDER = 'Hotmail/DBTraining'
     CLEAN_DESTINATION_FOLDER = True
     DISTORTION_W_MIN = 0
-    DISTORTION_W_MAX = 8
+    DISTORTION_W_MAX = 7
     DISTORTION_H_MIN = 0
-    DISTORTION_H_MAX = 8
+    DISTORTION_H_MAX = 7
     SCALE_MIN = 25
     SCALE_MAX = 29
     STEP = 1
@@ -90,11 +89,3 @@ def Myexcepthook(type, value, tb):
     Generate_Set(DESTINATION_FOLDER,CLEAN_DESTINATION_FOLDER,DISTORTION_W_MIN,DISTORTION_W_MAX,DISTORTION_H_MIN,
                  DISTORTION_H_MAX,SCALE_MIN,SCALE_MAX,STEP, elem_to_gen, FONTS, ALIGN_RANGEX, ALIGN_RANGEY, DEFAULT_SIZE)
 
-
-
-
-
-
-
-
-

Hotmail_4_GUI.py

@@ -1,15 +1,16 @@
-#!coding: utf-8
-
-MODEL_FOLDER = "Hotmail/Models"
-DEFAULT_MODEL_FILE="model.svm"
-
-CAPTCHA_FOLDER = "Hotmail/Rough Captchas"
-
-DEFAULT_CAPTCHA_FILE = "Image011.jpg"
-
-execfile("GUI.py")
-
-
-
-
-
+#!coding: utf-8
+#!coding: utf-8
+
+MODEL_FOLDER = "Hotmail/Models"
+DEFAULT_MODEL_FILE="model.svm"
+
+CAPTCHA_FOLDER = "Hotmail/Rough Captchas"
+
+DEFAULT_CAPTCHA_FILE = "Image011.jpg"
+
+execfile("GUI.py")
+
+
+
+
+

Hotmail_5_ComputeScores.py

@@ -156,7 +156,10 @@ def get_prediction(model, captcha, parent):
     for x in segs:
         for y in xrange(0, h):
             segmented_captcha.putpixel((x*parent.zoom, y), (255,0,0))
+
     parent.SetGraphImage(segmented_captcha)
+    parent.actif = False
+    parent.launchPredictionButton.SetLabel("Lancer la prédiction")
 
 
 

SVM_cross_validation.py

@@ -0,0 +1,68 @@
+#!coding: utf-8
+from svm import *
+import os, sys
+import Image
+
+import psyco
+psyco.full()
+
+from Preprocess import load_image
+from cross_validation import *
+
+
+#TRACEBACK
+import traceback
+import sys
+def Myexcepthook(type, value, tb):
+        lines=traceback.format_exception(type, value, tb)
+        f=open('log.txt', 'a')
+        f.write("\n".join(lines))
+        f.close()
+        print lines
+        raw_input()
+sys.excepthook=Myexcepthook
+
+
+print "Loading characters..."
+
+TRAINING_FOLDER = 'Egoshare/DBTraining-Captcha_based'
+
+labels = []
+samples = []
+
+for folder, subfolders, files in os.walk(TRAINING_FOLDER):
+    if folder[0] != ".":
+        loaded = False
+        for file in [file for file in files if 'bmp' in file]:
+            if not loaded:
+                print "Loading ", folder
+                loaded = True
+            im = Image.open(os.path.join(folder, file))
+            labels.append(ord(folder[-1])-65)
+            samples.append(map(lambda e:e/255., list(im.getdata())))
+print "Loading done."
+
+
+print "\nStarting cross-validation..."
+rates = []
+CRANGE = [1, 2, 4, 8, 16, 32, 64, 128, 256, 512, 1024, 2048, 4096, 8192, 16384, 32768, 65536]
+for C in CRANGE:
+    param = svm_parameter(kernel_type = RBF, C=C, probability = 1) #, gamma=1./(2*1.25**2))
+    rate = do_cross_validation(samples, labels, param, 10)
+    
+    f=open('cross-validation_results.txt', 'a')
+    f.write("C="+str(C)+"\tgamma=default\t=>\t"+str(rate)+'\n')
+    f.close()
+    
+    rates.append(rate)
+    
+
+print "Cross-validation done\n"
+print "SUCCES RATES: ", rates
+print "\nOPTIMAL PARAMETERS: "
+index = rates.index(max(rates))
+print "C = ", CRANGE[index]
+print "Optimal success rate: ", max(rates)
+
+raw_input()
+

cross-validation_results.txt

@@ -0,0 +1,11 @@
+HOTMAIL:
+--------
+#Cross-validation sur 580 caract�res
+C=1	gamma=default	=>	97.2222222222
+C=10	gamma=default	=>	99.1319444444
+C=100	gamma=default	=>	98.9583333333
+C=1000	gamma=default	=>	98.7847222222
+C=10000	gamma=default	=>	99.1319444444
+
+EGOSHARE (captcha-based):
+-------------------------