Recall eingebunden.

.gitignore

@@ -53,3 +53,5 @@ Thumbs.db
 
 assets/Bilder/*
 assets/Bilder/AktuelleTrainingsUndTestdaten/*
+assets/Bilder/Datengrundlage/*
+assets/Bilder/Datengrundlage/

main.py

@@ -21,10 +21,12 @@ def run_model():
     with open("ressources/config/config.yaml", "r") as file:
         config = yaml.load(file, Loader=yaml.FullLoader)
     # Initialisieren der KNN-Modellbau Eigenschaften
-    config_knn = ConfigKNN(excluded_folder=Markt.MARKT_B,
-                           activation_function_1_units=Aktivierungsfunktion.Sigmoid,
-                           activation_function_128_units=Aktivierungsfunktion.ReLU,
-                           optimization_method=Optimierungsverfahren.Adam)
+    config_knn = ConfigKNN(
+        excluded_folder=Markt.Kein_Markt,
+        activation_function_1_units=Aktivierungsfunktion.sigmoid,
+        activation_function_128_units=Aktivierungsfunktion.ReLU,
+        optimization_method=Optimierungsverfahren.SGD
+    )
 
     print("Initialisiere Callback")
     # Erstellen Sie eine Instanz des benutzerdefinierten Callbacks

ressources/config/config.yaml

 bilder:
   original_path: "/Users/dwillers/Programmierung/Master/Evaluation_OOS-Erkennung/assets/Bilder/Datengrundlage-Reduziert-Test/"
+  # original_path: "/Users/dwillers/Programmierung/Master/Evaluation_OOS-Erkennung/assets/Bilder/Datengrundlage/"
   knn_path: "/Users/dwillers/Programmierung/Master/Evaluation_OOS-Erkennung/assets/Bilder/AktuelleTrainingsUndTestdaten/"
 knn:
-  epochs: 2
+  epochs: 6
 result:
   plot_path: "/Users/dwillers/Programmierung/Master/Evaluation_OOS-Erkennung/ressources/results/plot/"
   excel_path: "/Users/dwillers/Programmierung/Master/Evaluation_OOS-Erkennung/ressources/results/excel/"

src/data/copyData.py

@@ -45,20 +45,21 @@ def copy_images_with_exclusion(src_dir, dest_dir, exclude_dir=None):
             os.remove(file_or_dir_path)
 
     # Erstelle die gewünschte Ordnerstruktur im Zielverzeichnis
-    os.makedirs(os.path.join(dest_dir, 'OOS'))
-    os.makedirs(os.path.join(dest_dir, '!OOS'))
+    os.makedirs(os.path.join(dest_dir, '0_OOS'))
+    os.makedirs(os.path.join(dest_dir, '1_!OOS'))
+
 
     # Durchlaufe die Unterordner im Quellverzeichnis
     for subdir in os.listdir(src_dir):
         #  if subdir == exclude_dir:
         subdirString = bytes(subdir, 'utf-8').decode('unicode_escape')
-        exclude_dirString = bytes(exclude_dir, 'utf-8').decode('unicode_escape')
+        exclude_dirString = bytes(exclude_dir.name, 'utf-8').decode('unicode_escape')
         if subdirString == exclude_dirString:
             continue
         for folder in os.listdir(os.path.join(src_dir, subdir)):
             if folder == 'OOS':
                 for img in os.listdir(os.path.join(src_dir, subdir, folder)):
-                    shutil.copy(os.path.join(src_dir, subdir, folder, img), os.path.join(dest_dir, 'OOS'))
+                    shutil.copy(os.path.join(src_dir, subdir, folder, img), os.path.join(dest_dir, '0_OOS'))
             elif folder == '!OOS':
                 for img in os.listdir(os.path.join(src_dir, subdir, folder)):
-                    shutil.copy(os.path.join(src_dir, subdir, folder, img), os.path.join(dest_dir, '!OOS'))
+                    shutil.copy(os.path.join(src_dir, subdir, folder, img), os.path.join(dest_dir, '1_!OOS'))

src/data/modelData.py

@@ -19,7 +19,7 @@ def define_augmentation_rules():
 def get_train_data(datagen, directory):
     train_it = datagen.flow_from_directory(
         directory=directory,
-        class_mode='binary',
+        class_mode='categorical',
         batch_size=64,
         target_size=(224, 224),
         subset='training')
@@ -29,7 +29,7 @@ def get_train_data(datagen, directory):
 def get_test_data(datagen, directory):
     test_it = datagen.flow_from_directory(
         directory=directory,
-        class_mode='binary',
+        class_mode='categorical',
         batch_size=64,
         target_size=(224, 224),
         subset='validation')

src/enum/activierungsfunktionEnum.py

@@ -3,4 +3,4 @@ from enum import Enum
 
 class Aktivierungsfunktion(Enum):
     ReLU = 0,
-    Sigmoid = 1,
+    sigmoid = 1,

src/enum/marktEnum.py

@@ -2,9 +2,10 @@ from enum import Enum
 
 
 class Markt(Enum):
-    MARKT_A = 'Markt_A',
-    MARKT_B = 'Markt_B',
-    MARKT_C = 'Markt_C',
-    MARKT_D = 'Markt_D',
-    MARKT_E = 'Markt_E',
-    MARKT_F = 'Markt_F'
+    Markt_A = 0,
+    Markt_B = 1,
+    Markt_C = 2,
+    Markt_D = 3,
+    Markt_E = 4,
+    Markt_F = 5
+    Kein_Markt = -1

src/knn/createKNN.py

@@ -12,5 +12,11 @@ def fit_model(config, model, train_it, test_it, callback):
 
 
 def evaluate_model(model, test_it):
-    _, acc = model.evaluate(test_it, steps=len(test_it), verbose=1)
-    print('> %.3f' % (acc * 100.0))
\ No newline at end of file
+    all_metrics = model.evaluate(test_it, steps=len(test_it), verbose=1)
+    _, metrics, recall_0, recall_1 = all_metrics
+    print(all_metrics)
+    print(_)
+    print(metrics)
+    print(recall_0)
+    print(recall_1)
+    print('> %.3f' % (metrics * 100.0))

src/knn/defineKNN.py

@@ -14,26 +14,31 @@ def define_model(config_knn):
     # Die Fully-Connected-Schichten werden definiert
     flat = tf.keras.layers.Flatten()(model.layers[-1].output)
     output_layer_1 = get_next_layer(128, config_knn.activation_function_128_units, flat)
-    output_layer_2 = get_next_layer(1, config_knn.activation_function_1_units, output_layer_1)
+    output_layer_2 = get_next_layer(2, config_knn.activation_function_1_units, output_layer_1)
+
     # Die Schichten des Modells werden definiert
     model = tf.keras.models.Model(inputs=model.inputs, outputs=output_layer_2)
 
     # Das Optimierungsverfahren wird definiert
     opt = get_optimization_method(config_knn.optimization_method)
     # Die Klassifizierungsmetriken und die Loss-Funktion werden definiert
-    recall_not_oos = tf.keras.metrics.Recall(name='recall_not_oos', class_id=0)
-    recall_oos = tf.keras.metrics.Recall(name='recall_oos', class_id=1)
-    model.compile(optimizer=opt, loss='binary_crossentropy', metrics=['accuracy', recall_not_oos, recall_oos])
+    loss_fn = tf.keras.losses.BinaryCrossentropy()
+    binary_accuracy = tf.keras.metrics.BinaryAccuracy()
+    recall_0 = tf.keras.metrics.Recall(name='recall_0', class_id=0)
+    recall_1 = tf.keras.metrics.Recall(name='recall_1', class_id=1)
+
+    # Füge die Metriken dem Modell hinzu
+    model.compile(optimizer=opt, loss=loss_fn, metrics=[binary_accuracy, recall_0, recall_1])
     return model
 
 
 def get_next_layer(units, activation_function, flat_or_dense):
     if activation_function == Aktivierungsfunktion.ReLU:
         return tf.keras.layers.Dense(units,
-                                     activation=activation_function,
+                                     activation=activation_function.name,
                                      kernel_initializer='he_uniform')(flat_or_dense)
-    if activation_function == Aktivierungsfunktion.Sigmoid:
-        return tf.keras.layers.Dense(units, activation='sigmoid')(flat_or_dense)
+    if activation_function == Aktivierungsfunktion.sigmoid:
+        return tf.keras.layers.Dense(units, activation=activation_function.name)(flat_or_dense)
     return None
 
 
@@ -43,3 +48,4 @@ def get_optimization_method(optimization_method):
     if optimization_method == Optimierungsverfahren.Adam:
         return tf.keras.optimizers.Adam(learning_rate=0.001)
     return None
+

src/result/createExcelFile.py

@@ -7,7 +7,19 @@ def create_excel_result(callback, config):
     workbook = openpyxl.Workbook()
 
     # Fügen Sie ein neues Arbeitsblatt hinzu
-    worksheet = workbook.create_sheet()
+    worksheet = workbook.active
+
+    # Setzen Sie die Spaltenbreite auf 19 Pixel
+    worksheet.column_dimensions["A"].width = 6
+    worksheet.column_dimensions["B"].width = 19
+    worksheet.column_dimensions["C"].width = 19
+    worksheet.column_dimensions["D"].width = 19
+    worksheet.column_dimensions["E"].width = 19
+    worksheet.column_dimensions["F"].width = 19
+    worksheet.column_dimensions["G"].width = 19
+    worksheet.column_dimensions["H"].width = 19
+    worksheet.column_dimensions["I"].width = 19
+    worksheet.column_dimensions["J"].width = 12
 
     # Basis Configurationen hier beschreiben
     # TODO: Implementieren der Excluded Folder, Aktivierungsfunktion (128 Einheiten, 1 Einheit), Optimierungsverfahren
@@ -15,22 +27,33 @@ def create_excel_result(callback, config):
     # Schreiben Sie die Überschriften in die erste Zeile
     # TODO: Ergänzung Recall 1, Recall 2
     worksheet.cell(row=1, column=1).value = 'epoch'
-    worksheet.cell(row=1, column=2).value = 'loss'
-    worksheet.cell(row=1, column=3).value = 'accuracy'
-    worksheet.cell(row=1, column=4).value = 'val_loss'
-    worksheet.cell(row=1, column=5).value = 'val_accuracy'
-    worksheet.cell(row=1, column=6).value = 'duration'
+    worksheet.cell(row=1, column=2).value = 'training_loss'
+    worksheet.cell(row=1, column=3).value = 'training_accuracy'
+    worksheet.cell(row=1, column=4).value = 'training_recall_not_oos'
+    worksheet.cell(row=1, column=5).value = 'training_recall_oos'
+    worksheet.cell(row=1, column=6).value = 'test_loss'
+    worksheet.cell(row=1, column=7).value = 'test_accuracy'
+    worksheet.cell(row=1, column=8).value = 'test_recall_not_oos'
+    worksheet.cell(row=1, column=9).value = 'test_recall_oos'
+    worksheet.cell(row=1, column=10).value = 'duration'
 
     # Schreiben Sie die Ergebnisse der Epochen in die Arbeitsmappe
-    for i, (loss, accuracy, val_loss, val_accuracy, duration) in enumerate(
-            zip(callback.epoch_losses, callback.epoch_accuracies, callback.epoch_val_losses,
-                callback.epoch_val_accuracies, callback.epoch_durations)):
+    for i, (loss, accuracy, recall_not_oos, recall_oos,
+            val_loss, val_accuracy, val_recall_not_oos, val_recall_oos, duration) in enumerate(
+            zip(callback.epoch_training_losses, callback.epoch_training_accuracies,
+                callback.epoch_training_recall_not_oos, callback.epoch_training_recall_oos,
+                callback.epoch_test_losses, callback.epoch_test_accuracies, callback.epoch_test_recall_not_oos,
+                callback.epoch_test_recall_oos, callback.epoch_durations)):
         worksheet.cell(row=i + 2, column=1).value = i + 1
         worksheet.cell(row=i + 2, column=2).value = loss
         worksheet.cell(row=i + 2, column=3).value = accuracy
-        worksheet.cell(row=i + 2, column=4).value = val_loss
-        worksheet.cell(row=i + 2, column=5).value = val_accuracy
-        worksheet.cell(row=i + 2, column=6).value = duration
+        worksheet.cell(row=i + 2, column=4).value = recall_not_oos
+        worksheet.cell(row=i + 2, column=5).value = recall_oos
+        worksheet.cell(row=i + 2, column=6).value = val_loss
+        worksheet.cell(row=i + 2, column=7).value = val_accuracy
+        worksheet.cell(row=i + 2, column=8).value = val_recall_not_oos
+        worksheet.cell(row=i + 2, column=9).value = val_recall_oos
+        worksheet.cell(row=i + 2, column=10).value = duration
 
     now = datetime.now()
     date_time = now.strftime("%d_%m_%y__%H_%M")

src/result/customCallback.py

@@ -3,19 +3,42 @@ import time
 
 
 class CustomCallback(tf.keras.callbacks.Callback):
-    def on_train_begin(self, logs={}):
-        self.epoch_losses = []
-        self.epoch_accuracies = []
-        self.epoch_val_losses = []
-        self.epoch_val_accuracies = []
+    def __init__(self):
+        super().__init__()
+        self.epoch_start_time = None
+        self.epoch_training_losses = None
+        self.epoch_training_accuracies = None
+        self.epoch_training_recall_oos = None
+        self.epoch_training_recall_not_oos = None
+        self.epoch_test_losses = None
+        self.epoch_test_accuracies = None
+        self.epoch_test_recall_oos = None
+        self.epoch_test_recall_not_oos = None
+        self.epoch_durations = None
+
+    def on_train_begin(self, logs=None):
+        self.epoch_training_losses = []
+        self.epoch_training_accuracies = []
+        self.epoch_training_recall_oos = []
+        self.epoch_training_recall_not_oos = []
+        self.epoch_test_losses = []
+        self.epoch_test_accuracies = []
+        self.epoch_test_recall_oos = []
+        self.epoch_test_recall_not_oos = []
         self.epoch_durations = []
 
-    def on_epoch_begin(self, epoch, logs={}):
+    def on_epoch_begin(self, epoch, logs=None):
         self.epoch_start_time = time.time()
 
-    def on_epoch_end(self, epoch, logs={}):
-        self.epoch_losses.append(logs.get('loss'))
-        self.epoch_accuracies.append(logs.get('accuracy'))
-        self.epoch_val_losses.append(logs.get('val_loss'))
-        self.epoch_val_accuracies.append(logs.get('val_accuracy'))
-        self.epoch_durations.append(time.time() - self.epoch_start_time)
\ No newline at end of file
+    def on_epoch_end(self, epoch, logs=None):
+        if logs is None:
+            logs = {}
+        self.epoch_training_losses.append(logs.get('loss'))
+        self.epoch_training_accuracies.append(logs.get('binary_accuracy'))
+        self.epoch_training_recall_oos.append(logs.get('recall_0'))
+        self.epoch_training_recall_not_oos.append(logs.get('recall_1'))
+        self.epoch_test_losses.append(logs.get('val_loss'))
+        self.epoch_test_accuracies.append(logs.get('val_binary_accuracy'))
+        self.epoch_test_recall_oos.append(logs.get('val_recall_0'))
+        self.epoch_test_recall_not_oos.append(logs.get('val_recall_1'))
+        self.epoch_durations.append(time.time() - self.epoch_start_time)

src/result/plotResult.py

@@ -14,8 +14,12 @@ def plot_values(history, config):
     # plot accuracy
     pyplot.subplot(212)
     pyplot.title('Classification Accuracy')
-    pyplot.plot(history.history['accuracy'], color='blue', label='train')
-    pyplot.plot(history.history['val_accuracy'], color='orange', label='test')
+    pyplot.plot(history.history['binary_accuracy'], color='blue', label='Accuracy')
+    pyplot.plot(history.history['recall_0'], color='blue', label='Recall OOS')
+    pyplot.plot(history.history['recall_1'], color='yellow', label='Recall !OOS')
+    pyplot.plot(history.history['val_binary_accuracy'], color='orange', label='Test Accuracy')
+    pyplot.plot(history.history['val_recall_0'], color='red', label='Test Recall OOS')
+    pyplot.plot(history.history['val_recall_1'], color='green', label='Test Recall !OOS')
     # save plot to file
     filename = sys.argv[0].split('/')[-1]
     pyplot.savefig(config["result"]["plot_path"] + filename + '_plot.png')