Dekker1/ResearchMethods
Archived
1
0
Fork 0
Code

merge

Browse Source
This commit is contained in:
Silver-T 2018-05-25 16:45:57 +10:00
commit 26803f8add
4 changed files with 197 additions and 99 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
mini_proj/_cutter.py
View File

@ -8,19 +8,19 @@ def image_cut(image, size0, size1):
dims = image.shape
assert dims[0] >= size0
assert dims[1] >= size1
return np.array([image[size0 * i:size0 * (i+1), size1 * j:size1 * (j+1)] \
return np.array([image[size0 * i:size0 * (i+1), size1 * j:size1 * (j+1), :] \
for i in range(dims[0] // size0) for j in range(dims[1] // size1)] + \
[image[size0 * i:size0 * (i+1), dims[1]-size1:] \
for i in range(dims[0] // size0) if dims[1] % size1 != 0] + \
[image[dims[0]-size0:, size1 * j:size1 * (j+1)] \
for j in range(dims[1] // size1) if dims[0] % size0 != \
for j in range(dims[1] // size1) if dims[0] % size0 != 0] \
)
if __name__ == '__main__':
# test = np.random.rand(5,4,3)
test = np.array([[
k + 4*j for k in range(4)
[k + 4*j, k + 4*j] for k in range(4)
] for j in range(5)])
print(test)
print(image_cut(test, 2, 2))

115
mini_proj/report/references.bib
View File

@ -1,9 +1,11 @@
@misc{openData,
title={Open Database License (ODbL) v1.0},
url={https://opendatacommons.org/licenses/odbl/1.0/},
journal={Open Data Commons},
year={2018},
month={Feb}
Classical Machine Learning
@article{MLReview,
title={Supervised machine learning: A review of classification techniques},
author={Kotsiantis, Sotiris B and Zaharakis, I and Pintelas, P},
journal={Emerging artificial intelligence applications in computer engineering},
volume={160},
pages={3--24},
year={2007}
}
@techreport{knn,
title={Discriminatory analysis-nonparametric discrimination: consistency properties},
@ -48,21 +50,52 @@
pages={18--22},
year={2002}
}
@article{Kotsiantis2007,
abstract = {Supervised machine learning is the search for algorithms that reason from externally supplied instances to produce general hypotheses, which then make predictions about future instances. In other words, the goal of supervised learning is to build a concise model of the distribution of class labels in terms of predictor features. The resulting classifier is then used to assign class labels to the testing instances where the values of the predictor features are known, but the value of the class label is unknown. This paper describes various supervised machine learning classification techniques. Of course, a single article cannot be a complete review of all supervised machine learning classification algorithms (also known induction classification algorithms), yet we hope that the references cited will cover the major theoretical issues, guiding the researcher in interesting research directions and suggesting possible bias combinations that have yet to be explored.},
author = {Kotsiantis, Sotiris B.},
doi = {10.1115/1.1559160},
file = {:home/kelvin/.local/share/data/Mendeley Ltd./Mendeley Desktop/Downloaded/Kotsiantis - 2007 - Supervised machine learning A review of classification techniques.pdf:pdf},
isbn = {1586037803},
issn = {09226389},
journal = {Informatica},
keywords = {algorithms analysis classifiers computational conn,classifiers,data mining techniques,intelligent data analysis,learning algorithms},
mendeley-groups = {CS Proj/ML,CS Proj,Thesis,Thesis/ML},
pages = {249--268},
title = {{Supervised machine learning: A review of classification techniques}},
url = {http://books.google.com/books?hl=en{\&}lr={\&}id=vLiTXDHr{\_}sYC{\&}oi=fnd{\&}pg=PA3{\&}dq=survey+machine+learning{\&}ots=CVsyuwYHjo{\&}sig=A6wYWvywU8XTc7Dzp8ZdKJaW7rc{\%}5Cnpapers://5e3e5e59-48a2-47c1-b6b1-a778137d3ec1/Paper/p800{\%}5Cnhttp://www.informatica.si/PDF/31-3/11{\_}Kotsiantis - S},
volume = {31},
year = {2007}
Neural Networks
@article{lenet,
title={Gradient-based learning applied to document recognition},
author={LeCun, Yann and Bottou, L{\'e}on and Bengio, Yoshua and Haffner, Patrick},
journal={Proceedings of the IEEE},
volume={86},
number={11},
pages={2278--2324},
year={1998},
publisher={IEEE}
}
@inproceedings{alexnet,
title={Imagenet classification with deep convolutional neural networks},
author={Krizhevsky, Alex and Sutskever, Ilya and Hinton, Geoffrey E},
booktitle={Advances in neural information processing systems},
pages={1097--1105},
year={2012}
}
@inproceedings{lenetVSalexnet,
title={On the Performance of GoogLeNet and AlexNet Applied to Sketches.},
author={Ballester, Pedro and de Ara{\'u}jo, Ricardo Matsumura},
booktitle={AAAI},
pages={1124--1128},
year={2016}
}
@article{deepNN,
title = "A survey of deep neural network architectures and their applications",
journal = "Neurocomputing",
volume = "234",
pages = "11 - 26",
year = "2017",
issn = "0925-2312",
doi = "https://doi.org/10.1016/j.neucom.2016.12.038",
url = "http://www.sciencedirect.com/science/article/pii/S0925231216315533",
author = "Weibo Liu and Zidong Wang and Xiaohui Liu and Nianyin Zeng and Yurong Liu and Fuad E. Alsaadi",
keywords = "Autoencoder, Convolutional neural network, Deep learning, Deep belief network, Restricted Boltzmann machine"
}
MISC
@misc{openData,
title={Open Database License (ODbL) v1.0},
url={https://opendatacommons.org/licenses/odbl/1.0/},
journal={Open Data Commons},
year={2018},
month={Feb}
}
@incollection{NIPS2012_4824,
title = {ImageNet Classification with Deep Convolutional Neural Networks},
@ -74,15 +107,33 @@ year = {2012},
publisher = {Curran Associates, Inc.},
url = {http://papers.nips.cc/paper/4824-imagenet-classification-with-deep-convolutional-neural-networks.pdf}
}
@ARTICLE{726791,
author={Y. Lecun and L. Bottou and Y. Bengio and P. Haffner},
journal={Proceedings of the IEEE},
title={Gradient-based learning applied to document recognition},
year={1998},
volume={86},
number={11},
pages={2278-2324},
keywords={backpropagation;convolution;multilayer perceptrons;optical character recognition;2D shape variability;GTN;back-propagation;cheque reading;complex decision surface synthesis;convolutional neural network character recognizers;document recognition;document recognition systems;field extraction;gradient based learning technique;gradient-based learning;graph transformer networks;handwritten character recognition;handwritten digit recognition task;high-dimensional patterns;language modeling;multilayer neural networks;multimodule systems;performance measure minimization;segmentation recognition;Character recognition;Feature extraction;Hidden Markov models;Machine learning;Multi-layer neural network;Neural networks;Optical character recognition software;Optical computing;Pattern recognition;Principal component analysis},
doi={10.1109/5.726791},
ISSN={0018-9219},
@ARTICLE{726791,
author={Y. Lecun and L. Bottou and Y. Bengio and P. Haffner},
journal={Proceedings of the IEEE},
title={Gradient-based learning applied to document recognition},
year={1998},
volume={86},
number={11},
pages={2278-2324},
keywords={backpropagation;convolution;multilayer perceptrons;optical character recognition;2D shape variability;GTN;back-propagation;cheque reading;complex decision surface synthesis;convolutional neural network character recognizers;document recognition;document recognition systems;field extraction;gradient based learning technique;gradient-based learning;graph transformer networks;handwritten character recognition;handwritten digit recognition task;high-dimensional patterns;language modeling;multilayer neural networks;multimodule systems;performance measure minimization;segmentation recognition;Character recognition;Feature extraction;Hidden Markov models;Machine learning;Multi-layer neural network;Neural networks;Optical character recognition software;Optical computing;Pattern recognition;Principal component analysis},
doi={10.1109/5.726791},
ISSN={0018-9219},
month={Nov},}
@book{numpy,
title={A guide to NumPy},
author={Oliphant, Travis E},
volume={1},
year={2006},
publisher={Trelgol Publishing USA}
}
@article{scikit-learn,
title={Scikit-learn: Machine Learning in {P}ython},
author={Pedregosa, F. and Varoquaux, G. and Gramfort, A. and Michel, V.
and Thirion, B. and Grisel, O. and Blondel, M. and Prettenhofer, P.
and Weiss, R. and Dubourg, V. and Vanderplas, J. and Passos, A. and
Cournapeau, D. and Brucher, M. and Perrot, M. and Duchesnay, E.},
journal={Journal of Machine Learning Research},
volume={12},
pages={2825--2830},
year={2011}
}

150
mini_proj/report/waldo.tex
View File

@ -30,7 +30,7 @@
\maketitle
\begin{abstract}
%
The famous brand of picture puzzles ``Where's Waldo?'' relates well to many
unsolved image classification problem. This offers us the opportunity to
test different image classification methods on a data set that is both small
@ -42,7 +42,7 @@
\todo{I don't like this big summation but I think it is the important
information}
Our comparison shows that \todo{...}
%
\end{abstract}
\section{Introduction}
@ -61,7 +61,7 @@
as) Waldo, but are not actually Waldo.
\begin{figure}[ht]
\includegraphics[scale=0.35]{waldo}
\includegraphics[scale=0.35]{waldo.png}
\centering
\caption{
A headshot of the character ``Waldo'', or ``Wally''. Pictures of Waldo
@ -106,7 +106,7 @@
The following paragraphs will give only brief descriptions of the different
classical machine learning methods used in this reports. For further reading
we recommend reading ``Supervised machine learning: A review of
classification techniques'' \cite{Kotsiantis2007}.
classification techniques'' \cite{MLReview}.
\paragraph{Naive Bayes Classifier}
@ -151,9 +151,11 @@
(binary) tree. Each non-leaf node contain a selection criteria to its
branches. Every leaf node contains the class that will be assigned to the
instance if the node is reached. In other training methods, decision trees
have the tendency to overfit, but in random forest a multitude of decision
tree is trained with a certain degree of randomness and the mean of these
trees is used which avoids this problem.
have the tendency to overfit\footnote{Overfitting occurs when a model learns
from the data too specifically, and loses its ability to generalise its
predictions for new data (resulting in loss of prediction accuracy)}, but in
random forest a multitude of decision tree is trained with a certain degree
of randomness and the mean of these trees is used which avoids this problem.
\subsection{Neural Network Architectures}
\tab There are many well established architectures for Neural Networks depending on the task being performed.
@ -168,43 +170,72 @@
\todo{Insert image of LeNet from slides if time}
\section{Method} \label{sec:method}
\tab
In order to effectively utilize the aforementioned modelling and classification techniques, a key consideration is the data they are acting on.
A dataset containing Waldo and non-Waldo images was obtained from an Open Database\footnote{``The Open Database License (ODbL) is a license agreement intended to allow users to freely share, modify, and use [a] Database while maintaining [the] same freedom for others"\cite{openData}}hosted on the predictive modelling and analytics competition framework, Kaggle.
The distinction between images containing Waldo, and those that do not, was providied by the separation of the images in different sub-directories.
It was therefore necessary to preprocess these images before they could be utilised by the proposed machine learning algorithms.
In order to effectively utilize the aforementioned modeling and
classification techniques, a key consideration is the data they are acting
on. A dataset containing Waldo and non-Waldo images was obtained from an
Open Database\footnote{``The Open Database License (ODbL) is a license
agreement intended to allow users to freely share, modify, and use [a]
Database while maintaining [the] same freedom for
others"\cite{openData}}hosted on the predictive modeling and analytics
competition framework, Kaggle. The distinction between images containing
Waldo, and those that do not, was provided by the separation of the images
in different sub-directories. It was therefore necessary to preprocess these
images before they could be utilized by the proposed machine learning
algorithms.
\subsection{Image Processing} \label{imageProcessing}
\tab
The Waldo image database consists of images of size 64$\times$64, 128$\times$128, and 256$\times$256 pixels obtained by dividing complete Where's Waldo? puzzles.
Within each set of images, those containing Waldo are located in a folder called `waldo', and those not containing Waldo, in a folder called `not\_waldo'.
Since Where's Waldo? puzzles are usually densely populated and contain fine details, the 64$\times$64 pixel set of images were selected to train and evaluate the machine learning models.
These images provide the added benefit of containing the most individual images of the three size groups.
\\
\par
Each of the 64$\times$64 pixel images were inserted into a Numpy
\footnote{Numpy is a popular Python programming library for scientific computing}
array of images, and a binary value was inserted into a seperate list at the same index.
These binary values form the labels for each image (waldo or not waldo).
Colour normalisation was performed on each so that artefacts in an image's colour profile correspond to meaningful features of the image (rather than photographic method).
\\
\par
Each original puzzle is broken down into many images, and only contains one Waldo. Although Waldo might span multiple 64$\times$64 pixel squares, this means that the non-Waldo data far outnumbers the Waldo data.
To combat the bias introduced by the skewed data, all Waldo images were artificially augmented by performing random rotations, reflections, and introducing random noise in the image to produce news images.
In this way, each original Waldo image was used to produce an additional 10 variations of the image, inserted into the image array.
This provided more variation in the true positives of the data set and assists in the development of more robust methods by exposing each technique to variations of the image during the training phase.
\\
\par
Despite the additional data, there were still over ten times as many non-Waldo images than Waldo images.
Therefore, it was necessary to cull the no-Waldo data, so that there was an even split of Waldo and non-Waldo images, improving the representation of true positives in the image data set. Following preprocessing, the images (and associated labels) were divided into a training and a test set with a 3:1 split.
\\
The Waldo image database consists of images of size 64$\times$64,
128$\times$128, and 256$\times$256 pixels obtained by dividing complete
``Where's Waldo?'' puzzles. Within each set of images, those containing
Waldo are located in a folder called \texttt{waldo}, and those not containing
Waldo, in a folder called \texttt{not\_waldo}. Since ``Where's Waldo?''
puzzles are usually densely populated and contain fine details, the
64$\times$64 pixel set of images were selected to train and evaluate the
machine learning models. These images provide the added benefit of
containing the most individual images of the three size groups. \\
Each of the 64$\times$64 pixel images were inserted into a
Numpy~\cite{numpy} array of images, and a binary value was inserted into a
separate list at the same index. These binary values form the labels for
each image (``Waldo'' or ``not Waldo''). Color normalization was performed
on each so that artifacts in an image's color profile correspond to
meaningful features of the image (rather than photographic method).\\
Each original puzzle is broken down into many images, and only contains one
Waldo. Although Waldo might span multiple 64$\times$64 pixel squares, this
means that the ``non-Waldo'' data far outnumbers the ``Waldo'' data. To
combat the bias introduced by the skewed data, all Waldo images were
artificially augmented by performing random rotations, reflections, and
introducing random noise in the image to produce news images. In this way,
each original Waldo image was used to produce an additional 10 variations of
the image, inserted into the image array. This provided more variation in
the true positives of the data set and assists in the development of more
robust methods by exposing each technique to variations of the image during
the training phase. \\
Despite the additional data, there were still ten times more ``non-Waldo''
images than Waldo images. Therefore, it was necessary to cull the
``non-Waldo'' data, so that there was an even split of ``Waldo'' and
``non-Waldo'' images, improving the representation of true positives in the
image data set. Following preprocessing, the images (and associated labels)
were divided into a training and a test set with a 3:1 split. \\
\subsection{Neural Network Training}\label{nnTraining}
\tab The neural networks used to classify the images were supervised learning models; requiring training on a dataset of typical images.
Each network was trained using the preprocessed training dataset and labels, for 25 epochs (one forward and backward pass of all data) in batches of 150.
The number of epochs was chosen to maximise training time and prevent overfitting\footnote{Overfitting occurs when a model learns from the data too specifically, and loses its ability to generalise its predictions for new data (resulting in loss of prediction accuracy)} of the training data, given current model parameters.
The batch size is the number of images sent through each pass of the network. Using the entire dataset would train the network quickly, but decrease the network's ability to learn unique features from the data.
Passing one image at a time may allow the model to learn more about each image, however it would also increase the training time and risk of overfitting the data.
Therefore the batch size was chosen to maintain training accuracy while minimising training time.
The neural networks used to classify the images were supervised learning
models; requiring training on a dataset of typical images. Each network was
trained using the preprocessed training dataset and labels for 25 epochs
(one forward and backward pass of all data) in batches of 150. The number of
epochs was chosen to maximize training time and prevent overfitting of the
training data, given current model parameters. The batch size is the number
of images sent through each pass of the network. Using the entire dataset
would train the network quickly, but decrease the network's ability to learn
unique features from the data. Passing one image at a time may allow the
model to learn more about each image, however it would also increase the
training time and risk of overfitting the data. Therefore the batch size was
chosen to maintain training accuracy while minimizing training time.
\subsection{Neural Network Testing}\label{nnTesting}
\tab After training each network, a separate test set of images (and labels) was used to evaluate the models.
@ -214,9 +245,9 @@
\subsection{Benchmarking}\label{benchmarking}
In order to benchmark the Neural Networks, the performance of these
algorithms are evaluated against other Machine Learning algorithms. We
use Support Vector Machines, K-Nearest Neighbours (\(K=5\)), Gaussian
Naive Bayes and Random Forest classifiers, as provided in Scikit-Learn.
algorithms are evaluated against other Machine Learning algorithms. We use
Support Vector Machines, K-Nearest Neighbors (\(K=5\)), Naive Bayes and
Random Forest classifiers, as provided in Scikit-Learn~\cite{scikit-learn}.
\subsection{Performance Metrics}\label{performance-metrics}
@ -243,22 +274,19 @@
\end{itemize}
\emph{Accuracy} is a common performance metric used in Machine Learning,
however in classification problems where the training data is heavily
biased toward one category, sometimes a model will learn to optimize its
accuracy by classifying all instances as one category. I.e. the
classifier will classify all images that do not contain Waldo as not
containing Waldo, but will also classify all images containing Waldo as
not containing Waldo. Thus we use, other metrics to measure performance
as well.
\\
\par
\emph{Precision} returns the percentage of classifications of Waldo that
are actually Waldo. \emph{Recall} returns the percentage of Waldos that
were actually predicted as Waldo. In the case of a classifier that
classifies all things as Waldo, the recall would be 0. \emph{F1-Measure}
returns a combination of precision and recall that heavily penalises
classifiers that perform poorly in either precision or recall.
% Kelvin End
however in classification problems where the training data is heavily biased
toward one category, sometimes a model will learn to optimize its accuracy
by classifying all instances as one category. I.e. the classifier will
classify all images that do not contain Waldo as not containing Waldo, but
will also classify all images containing Waldo as not containing Waldo. Thus
we use, other metrics to measure performance as well. \\
\emph{Precision} returns the percentage of classifications of Waldo that are
actually Waldo. \emph{Recall} returns the percentage of Waldos that were
actually predicted as Waldo. In the case of a classifier that classifies all
things as Waldo, the recall would be 0. \emph{F1-Measure} returns a
combination of precision and recall that heavily penalizes classifiers that
perform poorly in either precision or recall.
\section{Results} \label{sec:results}
\tab The time taken to train each of the neural networks and traditional approaches was measured and recorded alongside their accuracy (evaluated using a separate test dataset) in Table \ref{tab:results}.

25
mini_proj/test_nn.py
View File

@ -3,6 +3,7 @@ from keras.models import Model, load_model
from keras.utils import to_categorical
import cv2
from skimage import color, exposure
from _cutter import image_cut
def man_result_check():
pred_y = np.load("predicted_results.npy")
@ -14,7 +15,7 @@ def man_result_check():
z = 0
for i in range(0, len(test_y)):
print(pred_y[i], test_y[i], file=f)
# Calculates correct predictions
# Calculates correct predictions
if pred_y[i][0] == test_y[i][0]:
z+=1
@ -25,7 +26,7 @@ def man_result_check():
Purpose:Loads a trained neural network model (using Keras) to classify an image
Input: path/to/trained_model
image [or] path/to/image [if from_file=True]
Returns:Boolean variable
Returns:Boolean variable
'''
def is_Wally(trained_model_path, image, from_file=False):
if from_file:
@ -50,5 +51,23 @@ def is_Wally(trained_model_path, image, from_file=False):
# Mark Wally image somehow (colour the border)
# Stitch original image back together
if __name__ == '__main__':
# Read image
image = cv2.imread("10.jpg")
# Split image
cuts = image_cut(image, 64, 64)
for i in range(len(cuts)):
# Transform block
hsv = color.rgb2hsv(cuts[i])
hsv[:, :, 2] = exposure.equalize_hist(hsv[:, :, 2])
block = color.hsv2rgb(hsv)
block = np.rollaxis(block, -1)
if is_Wally("Waldo.h5", block):
# if True:
# Border block
GREEN = [0, 255, 0]
cuts[i] = cv2.copyMakeBorder(cuts[i][1:61,1:61],2,2,2,2,cv2.BORDER_CONSTANT,value=GREEN)
is_Wally("Waldo.h5", image)
# Stitch image TODO!
# Show image
cv2.imwrite('output.png',image)