cnn2

Author: Lazy Programmer

cnn_class2/content/elephant.jpg

@@ -0,0 +1,14 @@
+A Neural Algorithm of Artistic Style
+https://arxiv.org/abs/1508.06576
+
+SSD: Single Shot MultiBox Detector
+https://arxiv.org/abs/1512.02325
+
+Very Deep Convolutional Networks for Large-Scale Image Recognition (VGG)
+https://arxiv.org/abs/1409.1556
+
+Deep Residual Learning for Image Recognition
+https://arxiv.org/abs/1512.03385
+
+Going Deeper with Convolutions (Inception)
+https://arxiv.org/abs/1409.4842

cnn_class2/content/sydney.jpg

@@ -0,0 +1,109 @@
+# https://deeplearningcourses.com/advanced-computer-vision
+# https://www.udemy.com/advanced-computer-vision
+
+from __future__ import print_function, division
+from builtins import range
+# Note: you may need to update your version of future
+# sudo pip install -U future
+
+from keras.models import Sequential
+from keras.layers import Dense, Activation, Conv2D, MaxPooling2D, Flatten, Dropout, BatchNormalization
+
+import matplotlib.pyplot as plt
+import pandas as pd
+import numpy as np
+
+
+# helper
+def y2indicator(Y):
+  N = len(Y)
+  K = len(set(Y))
+  I = np.zeros((N, K))
+  I[np.arange(N), Y] = 1
+  return I
+
+
+# get the data
+# https://www.kaggle.com/zalando-research/fashionmnist
+data = pd.read_csv('../large_files/fashionmnist/fashion-mnist_train.csv')
+data = data.as_matrix()
+np.random.shuffle(data)
+
+X = data[:, 1:].reshape(-1, 28, 28, 1) / 255.0
+Y = data[:, 0].astype(np.int32)
+
+# get shapes
+# N = len(Y)
+K = len(set(Y))
+
+# by default Keras wants one-hot encoded labels
+# there's another cost function we can use
+# where we can just pass in the integer labels directly
+# just like Tensorflow / Theano
+Y = y2indicator(Y)
+
+
+# the model will be a sequence of layers
+model = Sequential()
+
+
+# make the CNN
+# model.add(Input(shape=(28, 28, 1)))
+model.add(Conv2D(input_shape=(28, 28, 1), filters=32, kernel_size=(3, 3)))
+model.add(BatchNormalization())
+model.add(Activation('relu'))
+model.add(MaxPooling2D())
+
+model.add(Conv2D(filters=64, kernel_size=(3, 3)))
+model.add(BatchNormalization())
+model.add(Activation('relu'))
+model.add(MaxPooling2D())
+
+model.add(Conv2D(filters=128, kernel_size=(3, 3)))
+model.add(BatchNormalization())
+model.add(Activation('relu'))
+model.add(MaxPooling2D())
+
+model.add(Flatten())
+model.add(Dense(units=300))
+model.add(Activation('relu'))
+model.add(Dropout(0.2))
+model.add(Dense(units=K))
+model.add(Activation('softmax'))
+
+
+# list of losses: https://keras.io/losses/
+# list of optimizers: https://keras.io/optimizers/
+# list of metrics: https://keras.io/metrics/
+model.compile(
+  loss='categorical_crossentropy',
+  optimizer='adam',
+  metrics=['accuracy']
+)
+
+# note: multiple ways to choose a backend
+# either theano, tensorflow, or cntk
+# https://keras.io/backend/
+
+
+# gives us back a <keras.callbacks.History object at 0x112e61a90>
+r = model.fit(X, Y, validation_split=0.33, epochs=15, batch_size=32)
+print("Returned:", r)
+
+# print the available keys
+# should see: dict_keys(['val_loss', 'acc', 'loss', 'val_acc'])
+print(r.history.keys())
+
+# plot some data
+plt.plot(r.history['loss'], label='loss')
+plt.plot(r.history['val_loss'], label='val_loss')
+plt.legend()
+plt.show()
+
+# accuracies
+plt.plot(r.history['acc'], label='acc')
+plt.plot(r.history['val_acc'], label='val_acc')
+plt.legend()
+plt.show()
+
+

cnn_class2/extra_reading.txt

@@ -0,0 +1,104 @@
+# https://deeplearningcourses.com/advanced-computer-vision
+# https://www.udemy.com/advanced-computer-vision
+
+from __future__ import print_function, division
+from builtins import range
+# Note: you may need to update your version of future
+# sudo pip install -U future
+
+from keras.models import Sequential, Model
+from keras.layers import Dense, Activation, Conv2D, MaxPooling2D, Flatten, Dropout, BatchNormalization, Input
+
+import matplotlib.pyplot as plt
+import pandas as pd
+import numpy as np
+
+
+# helper
+def y2indicator(Y):
+  N = len(Y)
+  K = len(set(Y))
+  I = np.zeros((N, K))
+  I[np.arange(N), Y] = 1
+  return I
+
+
+# get the data
+# https://www.kaggle.com/zalando-research/fashionmnist
+data = pd.read_csv('../large_files/fashionmnist/fashion-mnist_train.csv')
+data = data.as_matrix()
+np.random.shuffle(data)
+
+X = data[:, 1:].reshape(-1, 28, 28, 1) / 255.0
+Y = data[:, 0].astype(np.int32)
+
+# get shapes
+# N = len(Y)
+K = len(set(Y))
+
+# by default Keras wants one-hot encoded labels
+# there's another cost function we can use
+# where we can just pass in the integer labels directly
+# just like Tensorflow / Theano
+Y = y2indicator(Y)
+
+
+
+
+# make the CNN
+i = Input(shape=(28, 28, 1))
+x = Conv2D(filters=32, kernel_size=(3, 3))(i)
+x = BatchNormalization()(x)
+x = Activation('relu')(x)
+x = MaxPooling2D()(x)
+
+x = Conv2D(filters=64, kernel_size=(3, 3))(x)
+x = BatchNormalization()(x)
+x = Activation('relu')(x)
+x = MaxPooling2D()(x)
+
+x = Flatten()(x)
+x = Dense(units=100)(x)
+x = Activation('relu')(x)
+x = Dropout(0.3)(x)
+x = Dense(units=K)(x)
+x = Activation('softmax')(x)
+
+model = Model(inputs=i, outputs=x)
+
+
+# list of losses: https://keras.io/losses/
+# list of optimizers: https://keras.io/optimizers/
+# list of metrics: https://keras.io/metrics/
+model.compile(
+  loss='categorical_crossentropy',
+  optimizer='adam',
+  metrics=['accuracy']
+)
+
+# note: multiple ways to choose a backend
+# either theano, tensorflow, or cntk
+# https://keras.io/backend/
+
+
+# gives us back a <keras.callbacks.History object at 0x112e61a90>
+r = model.fit(X, Y, validation_split=0.33, epochs=15, batch_size=32)
+print("Returned:", r)
+
+# print the available keys
+# should see: dict_keys(['val_loss', 'acc', 'loss', 'val_acc'])
+print(r.history.keys())
+
+# plot some data
+plt.plot(r.history['loss'], label='loss')
+plt.plot(r.history['val_loss'], label='val_loss')
+plt.legend()
+plt.show()
+
+# accuracies
+plt.plot(r.history['acc'], label='acc')
+plt.plot(r.history['val_acc'], label='val_acc')
+plt.legend()
+plt.show()
+
+

cnn_class2/fashion.py

@@ -0,0 +1,39 @@
+# https://deeplearningcourses.com/advanced-computer-vision
+# https://www.udemy.com/advanced-computer-vision
+import os
+
+def mkdir(p):
+  if not os.path.exists(p):
+    os.mkdir(p)
+
+def link(src, dst):
+  if not os.path.exists(dst):
+    os.symlink(src, dst, target_is_directory=True)
+
+mkdir('../large_files/fruits-360-small')
+
+
+classes = [
+  'Apple Golden 1',
+  'Avocado',
+  'Lemon',
+  'Mango',
+  'Kiwi',
+  'Banana',
+  'Strawberry',
+  'Raspberry'
+]
+
+train_path_from = os.path.abspath('../large_files/fruits-360/Training')
+valid_path_from = os.path.abspath('../large_files/fruits-360/Validation')
+
+train_path_to = os.path.abspath('../large_files/fruits-360-small/Training')
+valid_path_to = os.path.abspath('../large_files/fruits-360-small/Validation')
+
+mkdir(train_path_to)
+mkdir(valid_path_to)
+
+
+for c in classes:
+  link(train_path_from + '/' + c, train_path_to + '/' + c)
+  link(valid_path_from + '/' + c, valid_path_to + '/' + c)