The previous Tutorial #08 showed how to use the pre-trained Inception model on the CIFAR-10 data-set in so-called Transfer Learning. This tutorial shows how to use your own images.
For demonstration, we use a new data-set called Knifey-Spoony that contains thousands of images of cutlery knives, spoons and forks on a few different backgrounds. The training-set contains 4170 images and the test-set contains 530 images. The classes are named knifey, spoony and forky as a reference to The Simpsons.
The images in the knifey-spoony data-set were created from video-files using a small Python script that is run on Linux (it requires the avconv-program for conversion from videos to images). This allows you to easily create very large data-sets with thousands of images from just a few minutes of video recordings.
This tutorial builds on the previous tutorials so you should be familiar with Tutorial #08 on Transfer Learning, as well as earlier tutorials on how to build and train Neural Networks in TensorFlow.
The following chart shows how the data flows when using the Inception model for Transfer Learning. First we input and process an image with the Inception model. Just prior to the final classification layer of the Inception model, we save the so-called Transfer Values to a cache-file.
This is very similar to how it was done in Tutorial #08, except that we now use the Knifey-Spoony data-set instead of CIFAR-10, which means that we are now feeding jpeg-images into the Inception model instead of numpy-arrays with image-data.
When all the images in the new data-set have been processed through the Inception model and the resulting transfer-values saved to a cache file, then we can use those transfer-values as the input to another neural network. We will then train the second neural network using the classes from the new data-set, so the network learns how to classify images based on the transfer-values from the Inception model.
In this way, the Inception model is used to extract useful information from the images and another neural network is then used for the actual classification.
In [1]:
from IPython.display import Image, display
Image('images/09_transfer_learning_flowchart.png')
Out[1]:
In [2]:
%matplotlib inline
import matplotlib.pyplot as plt
import tensorflow as tf
import numpy as np
import time
from datetime import timedelta
import os
# Functions and classes for loading and using the Inception model.
import inception
# We use Pretty Tensor to define the new classifier.
import prettytensor as pt
This was developed using Python 3.5.2 (Anaconda) and TensorFlow version:
In [3]:
tf.__version__
Out[3]:
PrettyTensor version:
In [4]:
pt.__version__
Out[4]:
In [5]:
import knifey
The data dimensions have already been defined in the knifey module, so we just need to import the ones we need.
In [6]:
from knifey import num_classes
Set the directory for storing the data-set on your computer.
In [7]:
# knifey.data_dir = "data/knifey-spoony/"
Set the directory that will be used for cache-files in this tutorial.
In [8]:
data_dir = knifey.data_dir
The Knifey-Spoony data-set is about 22 MB and will be downloaded automatically if it is not located in the given path.
In [9]:
knifey.maybe_download_and_extract()
Now load the data-set. This scans the sub-directories for all *.jpg images and puts the filenames into two lists for the training-set and test-set. This does not actually load the images, which will be done when the transfer-values are being calculated further below.
The lists of filenames are cached to harddisk so we can be sure that they are ordered in the same way when reloading the data-set later. This is important so we know which image-file corresponds to which transfer-values.
In [10]:
dataset = knifey.load()
You can use your own images instead of loading the knifey-spoony data-set. You have to create a DataSet-object from the dataset.py module. The best way is to use the load_cached()-wrapper-function which automatically saves a cache-file with the lists of image-files, so you make sure that the ordering is consistent with the transfer-values created below.
The images must be organized in sub-directories for each of the classes. See the documentation in the dataset.py module for more details.
In [11]:
# This is the code you would run to load your own image-files.
# It has been commented out so it won't run now.
# from dataset import load_cached
# dataset = load_cached(cache_path='my_dataset_cache.pkl', in_dir='my_images/')
# num_classes = dataset.num_classes
Get the class-names.
In [12]:
class_names = dataset.class_names
class_names
Out[12]:
Get the training-set. This returns the file-paths for the images, the class-numbers as integers, and the class-numbers as One-Hot encoded arrays called labels.
In [13]:
image_paths_train, cls_train, labels_train = dataset.get_training_set()
Print the first image-path to see if it looks OK.
In [14]:
image_paths_train[0]
Out[14]:
Get the test-set.
In [15]:
image_paths_test, cls_test, labels_test = dataset.get_test_set()
Print the first image-path to see if it looks OK.
In [16]:
image_paths_test[0]
Out[16]:
The Knifey-Spoony data-set has now been loaded and consists of 4700 images and associated labels (i.e. classifications of the images). The data-set is split into 2 mutually exclusive sub-sets, the training-set and the test-set.
In [17]:
print("Size of:")
print("- Training-set:\t\t{}".format(len(image_paths_train)))
print("- Test-set:\t\t{}".format(len(image_paths_test)))
Function used to plot at most 9 images in a 3x3 grid, and writing the true and predicted classes below each image.
In [18]:
def plot_images(images, cls_true, cls_pred=None, smooth=True):
assert len(images) == len(cls_true)
# Create figure with sub-plots.
fig, axes = plt.subplots(3, 3)
# Adjust vertical spacing.
if cls_pred is None:
hspace = 0.3
else:
hspace = 0.6
fig.subplots_adjust(hspace=hspace, wspace=0.3)
# Interpolation type.
if smooth:
interpolation = 'spline16'
else:
interpolation = 'nearest'
for i, ax in enumerate(axes.flat):
# There may be less than 9 images, ensure it doesn't crash.
if i < len(images):
# Plot image.
ax.imshow(images[i],
interpolation=interpolation)
# Name of the true class.
cls_true_name = class_names[cls_true[i]]
# Show true and predicted classes.
if cls_pred is None:
xlabel = "True: {0}".format(cls_true_name)
else:
# Name of the predicted class.
cls_pred_name = class_names[cls_pred[i]]
xlabel = "True: {0}\nPred: {1}".format(cls_true_name, cls_pred_name)
# Show the classes as the label on the x-axis.
ax.set_xlabel(xlabel)
# Remove ticks from the plot.
ax.set_xticks([])
ax.set_yticks([])
# Ensure the plot is shown correctly with multiple plots
# in a single Notebook cell.
plt.show()
This data-set does not load the actual images, instead it has a list of the images in the training-set and another list for the images in the test-set. This helper-function loads some image-files.
In [19]:
from matplotlib.image import imread
def load_images(image_paths):
# Load the images from disk.
images = [imread(path) for path in image_paths]
# Convert to a numpy array and return it.
return np.asarray(images)
In [20]:
# Load the first images from the test-set.
images = load_images(image_paths=image_paths_test[0:9])
# Get the true classes for those images.
cls_true = cls_test[0:9]
# Plot the images and labels using our helper-function above.
plot_images(images=images, cls_true=cls_true, smooth=True)
The Inception model is downloaded from the internet. This is the default directory where you want to save the data-files. The directory will be created if it does not exist.
In [21]:
# inception.data_dir = 'inception/'
Download the data for the Inception model if it doesn't already exist in the directory. It is 85 MB.
See Tutorial #07 for more details.
In [22]:
inception.maybe_download()
Load the Inception model so it is ready for classifying images.
Note the deprecation warning, which might cause the program to fail in the future.
In [23]:
model = inception.Inception()
Import a helper-function for caching the transfer-values of the Inception model.
In [24]:
from inception import transfer_values_cache
Set the file-paths for the caches of the training-set and test-set.
In [25]:
file_path_cache_train = os.path.join(data_dir, 'inception-knifey-train.pkl')
file_path_cache_test = os.path.join(data_dir, 'inception-knifey-test.pkl')
In [26]:
print("Processing Inception transfer-values for training-images ...")
# If transfer-values have already been calculated then reload them,
# otherwise calculate them and save them to a cache-file.
transfer_values_train = transfer_values_cache(cache_path=file_path_cache_train,
image_paths=image_paths_train,
model=model)
In [27]:
print("Processing Inception transfer-values for test-images ...")
# If transfer-values have already been calculated then reload them,
# otherwise calculate them and save them to a cache-file.
transfer_values_test = transfer_values_cache(cache_path=file_path_cache_test,
image_paths=image_paths_test,
model=model)
Check the shape of the array with the transfer-values. There are 4170 images in the training-set and for each image there are 2048 transfer-values.
In [28]:
transfer_values_train.shape
Out[28]:
Similarly, there are 530 images in the test-set with 2048 transfer-values for each image.
In [29]:
transfer_values_test.shape
Out[29]:
In [30]:
def plot_transfer_values(i):
print("Input image:")
# Plot the i'th image from the test-set.
image = imread(image_paths_test[i])
plt.imshow(image, interpolation='spline16')
plt.show()
print("Transfer-values for the image using Inception model:")
# Transform the transfer-values into an image.
img = transfer_values_test[i]
img = img.reshape((32, 64))
# Plot the image for the transfer-values.
plt.imshow(img, interpolation='nearest', cmap='Reds')
plt.show()
In [31]:
plot_transfer_values(i=100)
In [32]:
plot_transfer_values(i=300)
Use Principal Component Analysis (PCA) from scikit-learn to reduce the array-lengths of the transfer-values from 2048 to 2 so they can be plotted.
In [33]:
from sklearn.decomposition import PCA
Create a new PCA-object and set the target array-length to 2.
In [34]:
pca = PCA(n_components=2)
It takes a while to compute the PCA. In this case the data-set is not so large, but otherwise you could select a smaller part of the training-set to speed up the computation.
In [35]:
# transfer_values = transfer_values_train[0:3000]
transfer_values = transfer_values_train
Get the class-numbers for the samples you selected.
In [36]:
# cls = cls_train[0:3000]
cls = cls_train
Check that the array has 4170 samples and 2048 transfer-values for each sample.
In [37]:
transfer_values.shape
Out[37]:
Use PCA to reduce the transfer-value arrays from 2048 to 2 elements.
In [38]:
transfer_values_reduced = pca.fit_transform(transfer_values)
Check that it is now an array with 4170 samples and 2 values per sample.
In [39]:
transfer_values_reduced.shape
Out[39]:
Helper-function for plotting the reduced transfer-values.
In [40]:
def plot_scatter(values, cls):
# Create a color-map with a different color for each class.
import matplotlib.cm as cm
cmap = cm.rainbow(np.linspace(0.0, 1.0, num_classes))
# Create an index with a random permutation to make a better plot.
idx = np.random.permutation(len(values))
# Get the color for each sample.
colors = cmap[cls[idx]]
# Extract the x- and y-values.
x = values[idx, 0]
y = values[idx, 1]
# Plot it.
plt.scatter(x, y, color=colors, alpha=0.5)
plt.show()
Plot the transfer-values that have been reduced using PCA. There are 3 different colors for the different classes in the Knifey-Spoony data-set. The colors have very large overlap. This may be because PCA cannot properly separate the transfer-values.
In [41]:
plot_scatter(transfer_values_reduced, cls=cls)
In [42]:
from sklearn.manifold import TSNE
Another method for doing dimensionality reduction is t-SNE. Unfortunately, t-SNE is very slow so we first use PCA to reduce the transfer-values from 2048 to 50 elements.
In [43]:
pca = PCA(n_components=50)
transfer_values_50d = pca.fit_transform(transfer_values)
Create a new t-SNE object for the final dimensionality reduction and set the target to 2-dim.
In [44]:
tsne = TSNE(n_components=2)
Perform the final reduction using t-SNE. The current implemenation of t-SNE in scikit-learn cannot handle data with many samples so this might crash if you use the full training-set.
In [45]:
transfer_values_reduced = tsne.fit_transform(transfer_values_50d)
Check that it is now an array with 4170 samples and 2 transfer-values per sample.
In [46]:
transfer_values_reduced.shape
Out[46]:
Plot the transfer-values that have been reduced to 2-dim using t-SNE, which shows better separation than the PCA-plot above.
This means the transfer-values from the Inception model appear to contain enough information to separate the Knifey-Spoony images into classes, although there is still some overlap so the separation is not perfect.
In [47]:
plot_scatter(transfer_values_reduced, cls=cls)
Now we will create another neural network in TensorFlow. This network will take as input the transfer-values from the Inception model and output the predicted classes for Knifey-Spoony images.
It is assumed that you are already familiar with how to build neural networks in TensorFlow, otherwise see e.g. Tutorial #03.
First we need the array-length for transfer-values which is stored as a variable in the object for the Inception model.
In [48]:
transfer_len = model.transfer_len
Now create a placeholder variable for inputting the transfer-values from the Inception model into the new network that we are building. The shape of this variable is [None, transfer_len] which means it takes an input array with an arbitrary number of samples as indicated by the keyword None and each sample has 2048 elements, equal to transfer_len.
In [49]:
x = tf.placeholder(tf.float32, shape=[None, transfer_len], name='x')
Create another placeholder variable for inputting the true class-label of each image. These are so-called One-Hot encoded arrays with 3 elements, one for each possible class in the data-set.
In [50]:
y_true = tf.placeholder(tf.float32, shape=[None, num_classes], name='y_true')
Calculate the true class as an integer. This could also be a placeholder variable.
In [51]:
y_true_cls = tf.argmax(y_true, dimension=1)
Create the neural network for doing the classification on the Knifey-Spoony data-set. This takes as input the transfer-values from the Inception model which will be fed into the placeholder variable x. The network outputs the predicted class in y_pred.
See Tutorial #03 for more details on how to use Pretty Tensor to construct neural networks.
In [52]:
# Wrap the transfer-values as a Pretty Tensor object.
x_pretty = pt.wrap(x)
with pt.defaults_scope(activation_fn=tf.nn.relu):
y_pred, loss = x_pretty.\
fully_connected(size=1024, name='layer_fc1').\
softmax_classifier(num_classes=num_classes, labels=y_true)
Create a variable for keeping track of the number of optimization iterations performed.
In [53]:
global_step = tf.Variable(initial_value=0,
name='global_step', trainable=False)
Method for optimizing the new neural network.
In [54]:
optimizer = tf.train.AdamOptimizer(learning_rate=1e-4).minimize(loss, global_step)
The output of the network y_pred is an array with 3 elements. The class number is the index of the largest element in the array.
In [55]:
y_pred_cls = tf.argmax(y_pred, dimension=1)
Create an array of booleans whether the predicted class equals the true class of each image.
In [56]:
correct_prediction = tf.equal(y_pred_cls, y_true_cls)
The classification accuracy is calculated by first type-casting the array of booleans to floats, so that False becomes 0 and True becomes 1, and then taking the average of these numbers.
In [57]:
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
Once the TensorFlow graph has been created, we have to create a TensorFlow session which is used to execute the graph.
In [58]:
session = tf.Session()
The variables for the new network must be initialized before we start optimizing them.
In [59]:
session.run(tf.global_variables_initializer())
There are 4170 images (and arrays with transfer-values for the images) in the training-set. It takes a long time to calculate the gradient of the model using all these images (transfer-values). We therefore only use a small batch of images (transfer-values) in each iteration of the optimizer.
If your computer crashes or becomes very slow because you run out of RAM, then you may try and lower this number, but you may then need to perform more optimization iterations.
In [60]:
train_batch_size = 64
Function for selecting a random batch of transfer-values from the training-set.
In [61]:
def random_batch():
# Number of images (transfer-values) in the training-set.
num_images = len(transfer_values_train)
# Create a random index.
idx = np.random.choice(num_images,
size=train_batch_size,
replace=False)
# Use the random index to select random x and y-values.
# We use the transfer-values instead of images as x-values.
x_batch = transfer_values_train[idx]
y_batch = labels_train[idx]
return x_batch, y_batch
This function performs a number of optimization iterations so as to gradually improve the variables of the neural network. In each iteration, a new batch of data is selected from the training-set and then TensorFlow executes the optimizer using those training samples. The progress is printed every 100 iterations.
In [62]:
def optimize(num_iterations):
# Start-time used for printing time-usage below.
start_time = time.time()
for i in range(num_iterations):
# Get a batch of training examples.
# x_batch now holds a batch of images (transfer-values) and
# y_true_batch are the true labels for those images.
x_batch, y_true_batch = random_batch()
# Put the batch into a dict with the proper names
# for placeholder variables in the TensorFlow graph.
feed_dict_train = {x: x_batch,
y_true: y_true_batch}
# Run the optimizer using this batch of training data.
# TensorFlow assigns the variables in feed_dict_train
# to the placeholder variables and then runs the optimizer.
# We also want to retrieve the global_step counter.
i_global, _ = session.run([global_step, optimizer],
feed_dict=feed_dict_train)
# Print status to screen every 100 iterations (and last).
if (i_global % 100 == 0) or (i == num_iterations - 1):
# Calculate the accuracy on the training-batch.
batch_acc = session.run(accuracy,
feed_dict=feed_dict_train)
# Print status.
msg = "Global Step: {0:>6}, Training Batch Accuracy: {1:>6.1%}"
print(msg.format(i_global, batch_acc))
# Ending time.
end_time = time.time()
# Difference between start and end-times.
time_dif = end_time - start_time
# Print the time-usage.
print("Time usage: " + str(timedelta(seconds=int(round(time_dif)))))
Function for plotting examples of images from the test-set that have been mis-classified.
In [63]:
def plot_example_errors(cls_pred, correct):
# This function is called from print_test_accuracy() below.
# cls_pred is an array of the predicted class-number for
# all images in the test-set.
# correct is a boolean array whether the predicted class
# is equal to the true class for each image in the test-set.
# Negate the boolean array.
incorrect = (correct == False)
# Get the indices for the incorrectly classified images.
idx = np.flatnonzero(incorrect)
# Number of images to select, max 9.
n = min(len(idx), 9)
# Randomize and select n indices.
idx = np.random.choice(idx,
size=n,
replace=False)
# Get the predicted classes for those images.
cls_pred = cls_pred[idx]
# Get the true classes for those images.
cls_true = cls_test[idx]
# Load the corresponding images from the test-set.
# Note: We cannot do image_paths_test[idx] on lists of strings.
image_paths = [image_paths_test[i] for i in idx]
images = load_images(image_paths)
# Plot the images.
plot_images(images=images,
cls_true=cls_true,
cls_pred=cls_pred)
In [64]:
# Import a function from sklearn to calculate the confusion-matrix.
from sklearn.metrics import confusion_matrix
def plot_confusion_matrix(cls_pred):
# This is called from print_test_accuracy() below.
# cls_pred is an array of the predicted class-number for
# all images in the test-set.
# Get the confusion matrix using sklearn.
cm = confusion_matrix(y_true=cls_test, # True class for test-set.
y_pred=cls_pred) # Predicted class.
# Print the confusion matrix as text.
for i in range(num_classes):
# Append the class-name to each line.
class_name = "({}) {}".format(i, class_names[i])
print(cm[i, :], class_name)
# Print the class-numbers for easy reference.
class_numbers = [" ({0})".format(i) for i in range(num_classes)]
print("".join(class_numbers))
This function calculates the predicted classes of images and also returns a boolean array whether the classification of each image is correct.
The calculation is done in batches because it might use too much RAM otherwise. If your computer crashes then you can try and lower the batch-size.
In [65]:
# Split the data-set in batches of this size to limit RAM usage.
batch_size = 256
def predict_cls(transfer_values, labels, cls_true):
# Number of images.
num_images = len(transfer_values)
# Allocate an array for the predicted classes which
# will be calculated in batches and filled into this array.
cls_pred = np.zeros(shape=num_images, dtype=np.int)
# Now calculate the predicted classes for the batches.
# We will just iterate through all the batches.
# There might be a more clever and Pythonic way of doing this.
# The starting index for the next batch is denoted i.
i = 0
while i < num_images:
# The ending index for the next batch is denoted j.
j = min(i + batch_size, num_images)
# Create a feed-dict with the images and labels
# between index i and j.
feed_dict = {x: transfer_values[i:j],
y_true: labels[i:j]}
# Calculate the predicted class using TensorFlow.
cls_pred[i:j] = session.run(y_pred_cls, feed_dict=feed_dict)
# Set the start-index for the next batch to the
# end-index of the current batch.
i = j
# Create a boolean array whether each image is correctly classified.
correct = (cls_true == cls_pred)
return correct, cls_pred
Calculate the predicted class for the test-set.
In [66]:
def predict_cls_test():
return predict_cls(transfer_values = transfer_values_test,
labels = labels_test,
cls_true = cls_test)
This function calculates the classification accuracy given a boolean array whether each image was correctly classified. E.g. classification_accuracy([True, True, False, False, False]) = 2/5 = 0.4. The function also returns the number of correct classifications.
In [67]:
def classification_accuracy(correct):
# When averaging a boolean array, False means 0 and True means 1.
# So we are calculating: number of True / len(correct) which is
# the same as the classification accuracy.
# Return the classification accuracy
# and the number of correct classifications.
return correct.mean(), correct.sum()
Function for printing the classification accuracy on the test-set.
It takes a while to compute the classification for all the images in the test-set, that's why the results are re-used by calling the above functions directly from this function, so the classifications don't have to be recalculated by each function.
In [68]:
def print_test_accuracy(show_example_errors=False,
show_confusion_matrix=False):
# For all the images in the test-set,
# calculate the predicted classes and whether they are correct.
correct, cls_pred = predict_cls_test()
# Classification accuracy and the number of correct classifications.
acc, num_correct = classification_accuracy(correct)
# Number of images being classified.
num_images = len(correct)
# Print the accuracy.
msg = "Accuracy on Test-Set: {0:.1%} ({1} / {2})"
print(msg.format(acc, num_correct, num_images))
# Plot some examples of mis-classifications, if desired.
if show_example_errors:
print("Example errors:")
plot_example_errors(cls_pred=cls_pred, correct=correct)
# Plot the confusion matrix, if desired.
if show_confusion_matrix:
print("Confusion Matrix:")
plot_confusion_matrix(cls_pred=cls_pred)
The classification accuracy on the test-set is very low because the model variables have only been initialized and not optimized at all, so it just classifies the images randomly.
In [69]:
print_test_accuracy(show_example_errors=False,
show_confusion_matrix=True)
After 1000 optimization iterations, the classification accuracy is about 70% on the test-set.
In [70]:
optimize(num_iterations=1000)
In [71]:
print_test_accuracy(show_example_errors=True,
show_confusion_matrix=True)
We are now done using TensorFlow, so we close the session to release its resources. Note that we have two TensorFlow-sessions so we close both.
In [72]:
# This has been commented out in case you want to modify and experiment
# with the Notebook without having to restart it.
# model.close()
# session.close()
This tutorial showed how to use your own images in Transfer Learning with the Inception model. Thousands of images were used in this tutorial which were generated from just a few minutes of video recordings using a Python script.
However, the classification accuracy was not very good on the Knifey-Spoony data-set, especially for images of forks. The reason for this may be that the Inception model was originally trained on the ImageNet data-set which only contains 16 images of forks, while it contains more than 1200 images of spoons and more than 1300 images of knives. So it is possible that the Inception model cannot properly recognize forks.
We therefore need another technique for fine-tuning the Inception model so it becomes better at recognizing images of forks.
These are a few suggestions for exercises that may help improve your skills with TensorFlow. It is important to get hands-on experience with TensorFlow in order to learn how to use it properly.
You may want to backup this Notebook and the other files before making any changes.
*.pkl filenames and re-run this Notebook. Does this improve the classification accuracy? Compare the confusion matrix before and after this change.convert.py script. For example, make video recordings of cars and motorcycles and build a classification system that can tell those two classes apart.Copyright (c) 2016 by Magnus Erik Hvass Pedersen
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.