In this notebook, a template is provided for you to implement your functionality in stages which is required to successfully complete this project. If additional code is required that cannot be included in the notebook, be sure that the Python code is successfully imported and included in your submission, if necessary. Sections that begin with 'Implementation' in the header indicate where you should begin your implementation for your project. Note that some sections of implementation are optional, and will be marked with 'Optional' in the header.
In addition to implementing code, there will be questions that you must answer which relate to the project and your implementation. Each section where you will answer a question is preceded by a 'Question' header. Carefully read each question and provide thorough answers in the following text boxes that begin with 'Answer:'. Your project submission will be evaluated based on your answers to each of the questions and the implementation you provide.
Note: Code and Markdown cells can be executed using the Shift + Enter keyboard shortcut. In addition, Markdown cells can be edited by typically double-clicking the cell to enter edit mode.
In [1]:
# Load pickled data
import pickle
import pandas as pd
import numpy as np
training_file = 'train.p'
testing_file = 'test.p'
with open(training_file, mode='rb') as f:
train = pickle.load(f)
with open(testing_file, mode='rb') as f:
test = pickle.load(f)
X_train, y_train = train['features'], train['labels']
X_test, y_test = test['features'], test['labels']
assert(len(X_train) == len(y_train))
assert(len(X_test) == len(y_test))
# Load traffic sign names from CSV file
traffic_sign_names = pd.read_csv('signnames.csv', index_col='ClassId')
print()
print("Successfully completed data loading!")
The pickled data is a dictionary with 4 key/value pairs:
'features' is a 4D array containing raw pixel data of the traffic sign images, (num examples, width, height, channels).'labels' is a 2D array containing the label/class id of the traffic sign. The file signnames.csv contains id -> name mappings for each id.'sizes' is a list containing tuples, (width, height) representing the the original width and height the image.'coords' is a list containing tuples, (x1, y1, x2, y2) representing coordinates of a bounding box around the sign in the image. THESE COORDINATES ASSUME THE ORIGINAL IMAGE. THE PICKLED DATA CONTAINS RESIZED VERSIONS (32 by 32) OF THESE IMAGESComplete the basic data summary below.
In [2]:
# Number of training examples
n_train = len(X_train)
# Number of testing examples.
n_test = len(X_test)
# What's the shape of an traffic sign image?
image_shape = X_train[0].shape
# How many unique classes/labels there are in the dataset.
n_classes = np.unique(y_train).size
print("BASIC DATA SUMMARY")
print("==================")
print("Number of training examples =", n_train)
print("Number of testing examples =", n_test)
print("Image data shape =", image_shape)
print("Number of classes =", n_classes)
print()
print("TRAFFIC SIGN CLASSES")
print("====================")
traffic_sign_names
Out[2]:
Visualize the German Traffic Signs Dataset using the pickled file(s). This is open ended, suggestions include: plotting traffic sign images, plotting the count of each sign, etc.
The Matplotlib examples and gallery pages are a great resource for doing visualizations in Python.
NOTE: It's recommended you start with something simple first. If you wish to do more, come back to it after you've completed the rest of the sections.
In [3]:
### Data exploration visualization goes here.
import matplotlib.pyplot as plt
# Visualizations will be shown in the notebook.
%matplotlib inline
print("VISUALIZE ALL TRAFFIC SIGN CLASSES IN OUR DATASET")
print("=================================================")
cols = 11
rows = n_classes // cols + 1
fig = plt.figure(figsize=(cols, rows * 1.25))
for i in range(n_classes):
for j in range(len(y_train)):
if (i == y_train[j]):
plt.subplot(rows, cols, i+1)
plt.imshow(X_train[j])
plt.xlabel('{}'.format(i))
plt.xticks([])
plt.yticks([])
break
plt.show()
In [4]:
import collections
labels_count_train = collections.Counter(y_train)
labels_train, values_train = zip(*labels_count_train.items())
indexes_train = np.arange(len(labels_train))
labels_count_test = collections.Counter(y_test)
labels_test, values_test = zip(*labels_count_test.items())
indexes_test = np.arange(len(labels_test))
print("TRAINING AND TEST DATA DISTRIBUTION W.R.T TRAFFIC SIGN CLASSES")
print("==============================================================")
plt.subplot(211)
plt.bar(indexes_train, values_train, alpha=0.7)
plt.title('Training Data Distribution')
plt.xlabel("Traffic sign class")
plt.ylabel("Samples Count")
plt.subplot(212)
plt.bar(indexes_test, values_test, alpha=0.7, color='r')
plt.title('Test Data Distribution')
plt.xlabel("Traffic sign class")
plt.ylabel("Samples Count")
plt.tight_layout()
plt.show()
Design and implement a deep learning model that learns to recognize traffic signs. Train and test your model on the German Traffic Sign Dataset.
There are various aspects to consider when thinking about this problem:
Here is an example of a published baseline model on this problem. It's not required to be familiar with the approach used in the paper but, it's good practice to try to read papers like these.
NOTE: The LeNet-5 implementation shown in the classroom at the end of the CNN lesson is a solid starting point. You'll have to change the number of classes and possibly the preprocessing, but aside from that it's plug and play!
In [5]:
from scipy.ndimage import rotate
print("GENERATE ADDITIONAL DATA")
print('========================')
print('Generating additional training samples for traffic sign classes with less than 500 samples ...')
angles = [-2, 2, -5, 5, -7, 7, -10, 10]
# For each of the traffic sign class with less than 500 samples,
# add 3 of each samples with slight angle changes from -10 to 10 degrees
for i in range(len(labels_count_train)):
if labels_count_train[i] > 500:
continue
X_train_new = []
y_train_new = []
mask = np.where(y_train == i)
j = 0
for sample in X_train[mask]:
X_train_new.append(rotate(sample, angles[j % len(angles)], reshape=False))
y_train_new.append(i)
X_train_new.append(rotate(sample, angles[(j+1) % len(angles)], reshape=False))
y_train_new.append(i)
X_train_new.append(rotate(sample, angles[(j+2) % len(angles)], reshape=False))
y_train_new.append(i)
j += 1
X_train = np.append(X_train, X_train_new, axis=0)
y_train = np.append(y_train, y_train_new, axis=0)
# Number of training examples
n_train = len(X_train)
# Number of testing examples.
n_test = len(X_test)
# What's the shape of an traffic sign image?
image_shape = X_train[0].shape
# How many unique classes/labels there are in the dataset.
n_classes = np.unique(y_train).size
print()
print('DATA SUMMARY (After Additional Data)')
print('====================================')
print("Number of training examples =", n_train)
print("Number of testing examples =", n_test)
print("Image data shape =", image_shape)
print("Number of classes =", n_classes)
print('Done')
In [6]:
print("TRAINING DATA DISTRIBUTION BEFORE AND AFTER ADDITIONAL DATA")
print("===========================================================")
labels_count_after = collections.Counter(y_train)
labels_after, values_after = zip(*labels_count_after.items())
indexes_after = np.arange(len(labels_after))
plt.subplot(211)
plt.bar(indexes_train, values_train, alpha=0.7)
plt.title('Training Data Distribution (Before Additional Data)')
plt.xlabel("Traffic sign class")
plt.ylabel("Samples Count")
plt.subplot(212)
plt.bar(indexes_after, values_after, alpha=0.7, color='r')
plt.title('Training Data Distribution (After Additional Data)')
plt.xlabel("Traffic sign class")
plt.ylabel("Samples Count")
plt.tight_layout()
plt.show()
In [7]:
### Preprocess the data here.
### Feel free to use as many code cells as needed.
from sklearn.utils import shuffle
import random
from PIL import Image, ImageEnhance
print("PRE-PROCESS DATA")
print("================")
def enhanceImage(image_data, factor=2.0):
im = Image.fromarray(image_data, mode='RGB')
enhancer = ImageEnhance.Sharpness(im)
return np.array(enhancer.enhance(factor).getdata(),np.uint8).reshape(im.size[1], im.size[0], 3)
def normalize(image_data):
"""
Normalize the image data with Min-Max scaling to a range of [0.1, 0.9]
:param image_data: The image data to be normalized
:return: Normalized image data
"""
a = 0.1
b = 0.9
x_max = 255.0
x_min = 0.0
return a + (image_data * (b - a) / (x_max - x_min))
print("Normalize training and test samples ...")
X_train = normalize(X_train)
X_test = normalize(X_test)
print('Done')
Answer:
Answer:
In [8]:
### Define your architecture here.
### Feel free to use as many code cells as needed.
import tensorflow as tf
from tensorflow.contrib.layers import flatten
# reset graph
tf.reset_default_graph()
keep_prob = tf.placeholder(tf.float32, name='KeepProb')
def LeNet(x):
# Layer 1: Convolutional. Input = 32x32x3. Output = 32x32x16.
conv1_W = tf.get_variable("conv1_W", shape=[3, 3, 3, 16],
initializer=tf.contrib.layers.xavier_initializer())
conv1_b = tf.get_variable("conv1_b", shape=[16],
initializer=tf.contrib.layers.xavier_initializer())
conv1 = tf.nn.conv2d(x, conv1_W, strides=[1, 1, 1, 1], padding='SAME') + conv1_b
# Activation.
conv1 = tf.nn.relu(conv1)
# Dropout.
conv1 = tf.nn.dropout(conv1, keep_prob)
# Pooling. Input = 32x32x16. Output = 16x16x16.
conv1 = tf.nn.max_pool(conv1, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')
# Layer 2: Convolutional. Output = 16x16x64.
conv2_W = tf.get_variable("conv2_W", shape=[3, 3, 16, 64],
initializer=tf.contrib.layers.xavier_initializer())
conv2_b = tf.get_variable("conv2_b", shape=[64],
initializer=tf.contrib.layers.xavier_initializer())
conv2 = tf.nn.conv2d(conv1, conv2_W, strides=[1, 1, 1, 1], padding='SAME') + conv2_b
# Activation.
conv2 = tf.nn.relu(conv2)
# Dropout.
conv2 = tf.nn.dropout(conv2, keep_prob)
# Pooling. Input = 16x16x64. Output = 8x8x64.
conv2 = tf.nn.max_pool(conv2, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')
# Flatten. Input = 8x8x64. Output = 4096.
fc0 = flatten(conv2)
# Layer 3: Fully Connected. Input = 4096. Output = 512.
fc1_W = tf.get_variable("fc1_W", shape=[4096, 512],
initializer=tf.contrib.layers.xavier_initializer())
fc1_b = tf.get_variable("fc1_b", shape=[512],
initializer=tf.contrib.layers.xavier_initializer())
fc1 = tf.matmul(fc0, fc1_W) + fc1_b
# Activation.
fc1 = tf.nn.relu(fc1)
# Layer 4: Fully Connected. Input = 512. Output = 144.
fc2_W = tf.get_variable("fc2_W", shape=[512, 144],
initializer=tf.contrib.layers.xavier_initializer())
fc2_b = tf.get_variable("fc2_b", shape=[144],
initializer=tf.contrib.layers.xavier_initializer())
fc1 = tf.nn.dropout(fc1, keep_prob)
fc2 = tf.matmul(fc1, fc2_W) + fc2_b
# Activation.
fc2 = tf.nn.relu(fc2)
# Layer 5: Fully Connected. Input = 144. Output = 43.
fc3_W = tf.get_variable("fc3_W", shape=[144, 43],
initializer=tf.contrib.layers.xavier_initializer())
fc3_b = tf.get_variable("fc3_b", shape=[43],
initializer=tf.contrib.layers.xavier_initializer())
logits = tf.matmul(fc2, fc3_W) + fc3_b
return logits
What does your final architecture look like? (Type of model, layers, sizes, connectivity, etc.) For reference on how to build a deep neural network using TensorFlow, see Deep Neural Network in TensorFlow from the classroom.
Answer:
In [9]:
### Train your model here.
### Feel free to use as many code cells as needed.
EPOCHS = 25
BATCH_SIZE = 128
LEARNING_RATE = 0.001
LOG_PATH = './log'
# Split the train and validation, different dataset for each of the EPOCH
from sklearn.model_selection import StratifiedShuffleSplit
sss = StratifiedShuffleSplit(n_splits=EPOCHS, test_size=0.05, random_state=450)
X_train_org = X_train
y_train_org = y_train
x = tf.placeholder(tf.float32, (None, 32, 32, 3), name="InputData")
y = tf.placeholder(tf.int32, (None), name="LabelData")
one_hot_y = tf.one_hot(y, 43)
logits = LeNet(x)
with tf.name_scope('Model'):
prediction = tf.nn.softmax(logits)
with tf.name_scope('Loss'):
cross_entropy = tf.nn.softmax_cross_entropy_with_logits(logits, one_hot_y)
loss_operation = tf.reduce_mean(cross_entropy)
with tf.name_scope('Train'):
optimizer = tf.train.AdamOptimizer(learning_rate = LEARNING_RATE)
training_operation = optimizer.minimize(loss_operation)
with tf.name_scope('Accuracy'):
correct_prediction = tf.equal(tf.argmax(logits, 1), tf.argmax(one_hot_y, 1))
accuracy_operation = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
init = tf.global_variables_initializer()
# create a summary for our cost and accuracy
tf.summary.scalar("loss", loss_operation)
tf.summary.scalar("accuracy", accuracy_operation)
# merge all summaries into a single operation
summary_operation = tf.summary.merge_all()
saver = tf.train.Saver()
def evaluate(X_data, y_data):
num_examples = len(X_data)
total_accuracy = 0
total_loss = 0
sess = tf.get_default_session()
for offset in range(0, num_examples, BATCH_SIZE):
batch_x, batch_y = X_data[offset:offset+BATCH_SIZE], y_data[offset:offset+BATCH_SIZE]
accuracy, loss = sess.run([accuracy_operation, loss_operation] ,
feed_dict={x: batch_x, y: batch_y, keep_prob: 1.0})
total_accuracy += (accuracy * len(batch_x))
total_loss += (loss * len(batch_x))
return total_accuracy / num_examples, total_loss / num_examples
with tf.Session() as sess:
sess.run(init)
# Initialize logger for Tensorboard
writer = tf.summary.FileWriter(LOG_PATH, graph=tf.get_default_graph())
print("Training...")
print()
epoch = 0
for train_index, test_index in sss.split(X_train_org, y_train_org):
# Generate Training and Validation datasets for each of the EPOCH
X_train, X_validation = X_train_org[train_index], X_train_org[test_index]
y_train, y_validation = y_train_org[train_index], y_train_org[test_index]
num_examples = len(X_train)
for offset in range(0, num_examples, BATCH_SIZE):
end = offset + BATCH_SIZE
batch_x, batch_y = X_train[offset:end], y_train[offset:end]
_, summary = sess.run([training_operation, summary_operation], feed_dict={x: batch_x, y: batch_y, keep_prob: 0.5})
writer.add_summary(summary, epoch * BATCH_SIZE + offset)
validation_accuracy, validation_loss = evaluate(X_validation, y_validation)
print("EPOCH {} Validation Accuracy = {:.3f} Loss = {:.3f}".format(epoch+1, validation_accuracy, validation_loss))
epoch += 1
saver.save(sess, 'lenet')
print("Model saved")
What approach did you take in coming up with a solution to this problem? It may have been a process of trial and error, in which case, outline the steps you took to get to the final solution and why you chose those steps. Perhaps your solution involved an already well known implementation or architecture. In this case, discuss why you think this is suitable for the current problem.
Answer:
In [10]:
with tf.Session() as sess:
saver.restore(sess, tf.train.latest_checkpoint('.'))
test_accuracy, test_loss = evaluate(X_test, y_test)
print("Test accuracy of the model = {:.3f}".format(test_accuracy))
Take several pictures of traffic signs that you find on the web or around you (at least five), and run them through your classifier on your computer to produce example results. The classifier might not recognize some local signs but it could prove interesting nonetheless.
You may find signnames.csv useful as it contains mappings from the class id (integer) to the actual sign name.
In [11]:
from PIL import Image
print("SETTING UP NEW IMAGES")
print('=====================')
IMAGE_FOLDER = './new_images'
random_images = [
'35_ahead.jpg',
'14_stop_sign.png',
'5_speed_limit_80.png',
'22_bumpy_road.png',
'36_straight_or_right.png']
y_test_new = np.array([35,14,5,22,36])
cols = len(random_images)
rows = 1
fig = plt.figure(figsize=(cols, rows))
X_test_new = []
for idx, image_name in enumerate(random_images):
im = Image.open(IMAGE_FOLDER + '/' + image_name)
X_test_new.append(np.array(im.getdata(),
np.uint8).reshape(im.size[1], im.size[0], 3))
plt.subplot(rows, cols, idx+1)
plt.imshow(im)
plt.xlabel('{}'.format(y_test_new[idx]))
plt.xticks([])
plt.yticks([])
plt.show()
X_test_new = np.asarray(X_test_new, dtype=np.float32)
print("Normalizing new images ...")
X_test_new = normalize(X_test_new)
print('Done')
Answer:
Is your model able to perform equally well on captured pictures when compared to testing on the dataset? The simplest way to do this check the accuracy of the predictions. For example, if the model predicted 1 out of 5 signs correctly, it's 20% accurate.
NOTE: You could check the accuracy manually by using signnames.csv (same directory). This file has a mapping from the class id (0-42) to the corresponding sign name. So, you could take the class id the model outputs, lookup the name in signnames.csv and see if it matches the sign from the image.
In [12]:
### Run the predictions for the new images.
### Feel free to use as many code cells as needed.
print("PREDICTIONS FOR NEW IMAGES")
print("==========================")
print("Test accuracy for the model = {:.3f}".format(test_accuracy))
with tf.Session() as sess:
saver.restore(sess, tf.train.latest_checkpoint('.'))
new_images_accuracy = sess.run(accuracy_operation, feed_dict={x: X_test_new, y: y_test_new, keep_prob: 1.0})
new_images_prediction = sess.run(tf.argmax(tf.nn.softmax(logits),1), feed_dict={x: X_test_new, keep_prob: 1.0})
new_images_certainity = sess.run(tf.nn.top_k(logits, 3), feed_dict={x: X_test_new, keep_prob: 1.0})
print("Prediction accuracy for new images = {:.3f}".format(new_images_accuracy))
print("Expected result (E) =", y_test_new)
print("Predicted result (P) =", new_images_prediction)
print()
for i, (predictions, probabilities, expected_class) in enumerate(zip(new_images_certainity.indices, new_images_certainity.values, y_test_new)):
print("Expected class ({}) Predictions ({}) Probabilities ({})".format(expected_class, predictions, probabilities))
Use the model's softmax probabilities to visualize the certainty of its predictions, tf.nn.top_k could prove helpful here. Which predictions is the model certain of? Uncertain? If the model was incorrect in its initial prediction, does the correct prediction appear in the top k? (k should be 5 at most)
tf.nn.top_k will return the values and indices (class ids) of the top k predictions. So if k=3, for each sign, it'll return the 3 largest probabilities (out of a possible 43) and the correspoding class ids.
Take this numpy array as an example:
# (5, 6) array
a = np.array([[ 0.24879643, 0.07032244, 0.12641572, 0.34763842, 0.07893497,
0.12789202],
[ 0.28086119, 0.27569815, 0.08594638, 0.0178669 , 0.18063401,
0.15899337],
[ 0.26076848, 0.23664738, 0.08020603, 0.07001922, 0.1134371 ,
0.23892179],
[ 0.11943333, 0.29198961, 0.02605103, 0.26234032, 0.1351348 ,
0.16505091],
[ 0.09561176, 0.34396535, 0.0643941 , 0.16240774, 0.24206137,
0.09155967]])Running it through sess.run(tf.nn.top_k(tf.constant(a), k=3)) produces:
TopKV2(values=array([[ 0.34763842, 0.24879643, 0.12789202],
[ 0.28086119, 0.27569815, 0.18063401],
[ 0.26076848, 0.23892179, 0.23664738],
[ 0.29198961, 0.26234032, 0.16505091],
[ 0.34396535, 0.24206137, 0.16240774]]), indices=array([[3, 0, 5],
[0, 1, 4],
[0, 5, 1],
[1, 3, 5],
[1, 4, 3]], dtype=int32))Looking just at the first row we get [ 0.34763842, 0.24879643, 0.12789202], you can confirm these are the 3 largest probabilities in a. You'll also notice [3, 0, 5] are the corresponding indices.
Answer:
In [13]:
# plot softmax probabilities to visualize the certainty
for i, (predictions, probabilities, expected_image, expected_class) in enumerate(zip(new_images_certainity.indices, new_images_certainity.values, X_test_new, y_test_new)):
fig = plt.figure(figsize=(15, 2))
plt.bar(predictions, probabilities)
plt.ylabel('Certainity in %')
#plt.yticks(np.arange(0.0, 50.0, 10), np.arange(0.0, 50.0, 10))
plt.xticks(np.arange(0.5, 43.5, 1.0), list(traffic_sign_names.values.flatten()), ha='right', rotation=45)
ax = plt.axes([0.75,0.25,0.5,0.5], frameon=True)
ax.imshow(expected_image)
ax.set_xlabel('E:{} P:{}'.format(expected_class, predictions[0]))
ax.set_xticks([])
ax.set_yticks([])
plt.show()
Note: Once you have completed all of the code implementations and successfully answered each question above, you may finalize your work by exporting the iPython Notebook as an HTML document. You can do this by using the menu above and navigating to \n", "File -> Download as -> HTML (.html). Include the finished document along with this notebook as your submission.