ส่งภาพ ทำส่วน machine learning

sessions/session-4/ubuwidgets.py

@@ -62,14 +62,14 @@ class Board(QWidget):
         cnt = 0
         maxf = 0
         for file in os.listdir(label):
-            if file.endswith('.bmp'):
+            if file.endswith('.png'):
                 cnt += 1
                 tmpN = int(file[:-4:])
                 maxf = max(maxf,tmpN)
-        visibleImage.save('tmp.bmp')
-        pixmap = QPixmap('tmp.bmp').scaled(8,8)
-        pixmap.save(label+'/'+str(max(cnt,maxf)+1)+'.bmp')
-        QFile.remove('tmp.bmp')
+        visibleImage.save('tmp.png')
+        pixmap = QPixmap('tmp.png').scaled(64,64)
+        pixmap.save(label+'/'+str(max(cnt,maxf)+1)+'.png')
+        QFile.remove('tmp.png')
         
 
     def paintEvent(self, event):

sessions/session-4/window.py

@@ -5,6 +5,16 @@ from PyQt5.QtWidgets import (QAction, QApplication, QWidget,QGridLayout)
 #from trainWidget import TrainWidget
 #from testWidget import TestWidget
 from ubuwidgets import (Button, ComboWidget, Board)
+import os
+import matplotlib.pyplot as plt
+
+# Import datasets, classifiers and performance metrics
+from sklearn import datasets, svm, metrics
+from PIL import Image
+import numpy as np
+import math
+from sklearn.metrics import accuracy_score
+from sklearn.externals import joblib
 
 class MainWindow(QWidget):
     def __init__(self,parent=None):
@@ -26,9 +36,55 @@ class MainWindow(QWidget):
 
     def on_clear(self):
         self.scribbleArea.clearImage()
-
+        
+    def data_prep(self):
+        print('preporcessing..')
+        
     def on_train(self):
+        self.data_prep()
+        
         print('training..')
+        list_of_label = os.listdir('Image/')
+        digit_data = list()
+        digit_label = list()
+        for label in list_of_label:
+            label_path = 'Image/'+str(label)+'/'
+            list_of_data = os.listdir(label_path)
+           #print(list_of_data)
+            for data in list_of_data:
+                im = Image.open(label_path+'/'+data)
+                arr_im = np.array(im)
+                list_im = str(arr_im.tolist())
+                
+                width, height = im.size
+                gen_data = np.zeros(width*height)
+                count = 0
+                for i in range(len(arr_im)):
+                    for j in range(len(arr_im[i])):
+                        #ทำ Normalize ด้วยการ (R+G+B) หาร 765 มาจาก (255+255+255)
+                        gen_data[count] = math.floor(((np.sum(arr_im[i][j][:3])/765)))
+                        count+=1
+                digit_data.append( gen_data )
+                digit_label.append( label )     
+        digit_data = np.array(digit_data)  #ข้อมูล
+        digit_label = np.array(digit_label) # label ของข้อมูล  
+        
+        
+        # Create a classifier: a support vector classifier
+        classifier = svm.SVC(gamma=0.001, C=.1)
+        
+        # We learn the digits on the first half of the digits
+        classifier.fit(digit_data, digit_label)
+        #save model
+        joblib.dump(classifier, 'svmModel.pkl')
+        
+        expected = digit_label
+        predicted = classifier.predict(digit_data)
+        print("Classification report for classifier %s:\n%s\n"
+              % (classifier, metrics.classification_report(expected, predicted)))
+        print("Confusion matrix:\n%s" % metrics.confusion_matrix(expected, predicted))
+        print("\naccuracy = ",accuracy_score(digit_label, predicted))        
+        
         
     def on_save(self):
         print('on_save')
@@ -37,6 +93,21 @@ class MainWindow(QWidget):
         
     def on_test(self):
         print("testing...")
+        clf = joblib.load('svmModel.pkl')
+        
+        print('preporcessing..')
+        i = Image.open('test/t.png')
+        iar = np.array(i)
+        test_data = np.zeros(64*64)
+        len(iar[0])
+        count = 0
+        for i in range(len(iar)):
+            for j in range(len(iar[i])):
+                test_data[count] = math.floor(((np.sum(iar[i][j][:3])/765)))
+                #print(test_data[count])
+                count+=1
+        predicted = clf.predict([test_data])
+        print(predicted)
         
         
 if __name__ == '__main__':