refinement for the learning

network/Resnet50.py

-from network_helpers.NetworkBase import NetworkBase
+from network.NetworkBase import NetworkBase
 from tensorflow.keras.models import Model
 from tensorflow.keras.layers import Input, Dense, GlobalAveragePooling2D, Dropout, Flatten, AveragePooling2D
 from tensorflow.keras.applications.resnet_v2 import ResNet50V2
@@ -29,13 +29,16 @@ class Resnet50(NetworkBase):
         if shouldSave:
             base_model.save('input/resnet_50_base_model.h5')
 
+        for layer in base_model.layers:
+            layer.trainable = False
+
         x = GlobalAveragePooling2D()(base_model.output)
         x = Dropout(0.5)(x)
         out = Dense(self.NUM_CLASSES, activation='softmax', name='probs')(x)
 
         model = Model(base_model.input, out)
         
-        for layer in model.layers[:-39]:
-            layer.trainable = False
+        #for layer in model.layers[:-39]:
+        #    layer.trainable = False
 
         return model

train_classifier.py

@@ -2,12 +2,10 @@ from datetime import datetime
 from optparse import OptionParser
 
 from PIL import ImageFile
-from keras_preprocessing.image import ImageDataGenerator
-from network_helpers.Resnet152 import Resnet152
-from network_helpers.Resnet50 import Resnet50
-from network_helpers.DataLoader import DataLoader
-from tensorflow.keras.optimizers import Adam
+from network.Resnet50 import Resnet50
+from helpers.DataLoader import DataLoader
 from tensorflow.keras.callbacks import ModelCheckpoint, TensorBoard, EarlyStopping
+from tensorflow.keras.optimizers import RMSprop, SGD
 
 import pathlib
 
@@ -23,6 +21,7 @@ parser.add_option("-o", "--path_output", dest="output_path", help="Path to base
 parser.add_option("-t", "--path_test", dest="test_path", help="Path to test input.", default="./input/test_data")
 parser.add_option("--base_network_file", dest="base_net_file", help="Optional link to local file of Resnet 152 V2 for TF without top.")
 parser.add_option("--num_epochs", type="int", dest="num_epochs", help="Number of epochs.", default=100)
+parser.add_option("--num_epochs_pre_train", type="int", dest="num_epochs_pre_train", help="Number of epochs for the basic training of the FC layers.", default=50)
 parser.add_option("--batch_size", type="int", dest="batch_size", help="Size of batches.", default=10)
 parser.add_option("--patience", type="int", dest="early_stopping_patience", help="How many epochs without improvement should the training going on?", default=5)
 parser.add_option("--split_size", type="float", dest="split_size", help="Proportion of validation examples out of all test examples. Set as value between 0 and 1.", default=0.5)
@@ -92,12 +91,35 @@ model = network.get_model()
 
 #compile the model
 
-model.compile(optimizer='rmsprop', loss='categorical_crossentropy', metrics=['accuracy'])
-#model.compile(optimizer=Adam(lr=0.000001), loss='categorical_crossentropy', metrics=['accuracy'])
+model.compile(optimizer=RMSprop(), loss='categorical_crossentropy', metrics=['accuracy'])
 
 # Print Network summary
 # model.summary()
 
+'''
+    Pre-Train FC Layers 
+'''
+callbacks_prefit = [
+    TensorBoard(options.output_path +'/logs/{}_prefit'.format(time)),
+]
+
+history = model.fit(train_dataset,
+                    epochs=options.num_epochs_pre_train,
+                    validation_data=validation_dataset,
+                    callbacks=callbacks_prefit,
+                    verbose=2
+                    )
+#model.save_weights('last_pre_train_model.h5')
+print('\nHistory dict:', history.history)
+
+for layer in model.layers[-39:]:
+    layer.trainable = True
+
+model.compile(optimizer=SGD(lr=1e-4, momentum=0.9), loss='categorical_crossentropy', metrics=['accuracy'])
+
+# Reload Training Data to shuffle and augment in a different way then before
+train_dataset = train_loader.load_dataset()
+
 callbacks = [
     ModelCheckpoint(
         filepath= str(checkpoint_path) + '/model.{epoch:04d}-{val_loss:.3f}.hdf5',