top5accuracy

Author: ewwx

gierlach/build_model_top5.py

@@ -0,0 +1,75 @@
+import pandas as pd
+import numpy as np 
+import matplotlib.pyplot as plt
+import matplotlib.image as mpimg
+from keras.models import Sequential
+from keras.layers import Conv2D
+from keras.layers import MaxPooling2D
+from keras.preprocessing.image import ImageDataGenerator
+from keras.layers.core import Dense, Dropout, Activation, Flatten
+from keras.metrics import top_k_categorical_accuracy
+
+def top_5_accuracy(y_true, y_pred):
+    return top_k_categorical_accuracy(y_true, y_pred, k=5)
+
+def get_data():
+    train_data = ImageDataGenerator(rescale = 1./255, shear_range = 0.2,zoom_range = 0.2,rotation_range=40,fill_mode='nearest', horizontal_flip = True)
+    test_data = ImageDataGenerator(rescale = 1./255)
+    validation_data = ImageDataGenerator(rescale = 1./255)
+    training_set = train_data.flow_from_directory('Training',target_size = (64, 64),color_mode='rgb',class_mode = 'categorical')
+    test_set = test_data.flow_from_directory('Testing',target_size = (64,64),color_mode='rgb',class_mode = 'categorical')
+    validation_set = validation_data.flow_from_directory('Validation',target_size = (64,64),color_mode='rgb',class_mode = 'categorical')
+    plt.figure()
+    plt.hist(training_set.classes, bins=62) 
+    plt.title("Histogram of training dataset")
+    plt.savefig('Histogram of training dataset.png')
+    plt.figure()
+    plt.hist(test_set.classes, bins=62)  
+    plt.title("Histogram of testing dataset")
+    plt.savefig("Histogram of testing dataset.png")
+    plt.figure()
+    plt.hist(validation_set.classes, bins=62)  
+    plt.title("Histogram of validation dataset")
+    plt.savefig("Histogram of validation dataset.png")
+
+    classifier = Sequential()
+    classifier.add(Conv2D(32, (3, 3), input_shape = (64, 64, 3), activation = 'relu'))
+    classifier.add(Conv2D(32, (3, 3), input_shape = (64, 64, 3), activation = 'relu'))
+    classifier.add(MaxPooling2D(pool_size = (2, 2)))
+    classifier.add(Conv2D(32, (3, 3), activation = 'relu'))
+    classifier.add(MaxPooling2D(pool_size = (2, 2)))
+    classifier.add(Flatten(input_shape=classifier.output_shape[1:]))
+    classifier.add(Dense(256, activation='relu'))
+    classifier.add(Dropout(0.5))
+    classifier.add(Dense(62, activation='softmax'))
+    classifier.compile(optimizer = 'Nadam', loss = 'mean_squared_error', metrics = [top_5_accuracy])
+
+
+    nr_epoch = 20
+    his = classifier.fit_generator(training_set,steps_per_epoch = 110,epochs = nr_epoch,validation_data = validation_set)
+    train_loss = his.history['loss']
+    val_loss   = his.history['val_loss']
+    train_acc  = his.history['top_5_accuracy']
+    val_acc    = his.history['val_top_5_accuracy']
+    xc         = range(nr_epoch)
+
+    plt.figure()
+    plt.plot(xc, val_loss, label='val_loss')
+    plt.plot(xc, train_loss, label='train_loss')
+    plt.ylabel("mean_squared_error")
+    plt.xlabel("number of epoch")
+    plt.legend()
+    plt.savefig("Loss_top5.png")
+    plt.figure()
+    plt.plot(xc, val_acc, label='val_acc')
+    plt.plot(xc ,train_acc, label='train_acc')
+    plt.legend(loc=4)
+    plt.ylabel("accuracy")
+    plt.xlabel("number of epoch")
+    plt.savefig("Accuracy_top5.png")
+    classifier.save('classifier_20epoch.h5')
+  
+
+get_data()
+
+

gierlach/draw_model.py

@@ -0,0 +1,10 @@
+from keras.models import load_model
+from keras.utils import plot_model
+
+model = load_model('classifier_augmentation1_20epoch.h5')
+plot_model(model, to_file='model.png')
+
+
+
+
+