Thesis_draft0.2.lof

\babel@toc {british}{}\relax 
\addvspace {10\p@ }
\contentsline {figure}{\numberline {1.1}{\ignorespaces Data contribution of buildings and highway in all and humanitarian settings within OSM (Herfort et al., 2021)}}{10}{figure.1.1}%
\addvspace {10\p@ }
\contentsline {figure}{\numberline {2.1}{\ignorespaces The four main types of Computer Vision tasks (Stanford University, 2022)}}{15}{figure.2.1}%
\contentsline {figure}{\numberline {2.2}{\ignorespaces Schematic analogy diagram between a biological neuron and an artificial perceptron (Fumo D., 2017).}}{16}{figure.2.2}%
\contentsline {figure}{\numberline {2.3}{\ignorespaces Schematic diagram of a CNN (Stanford University, 2022).}}{17}{figure.2.3}%
\contentsline {figure}{\numberline {2.4}{\ignorespaces 3 x 3 Convolution (Stanford University, 2022).}}{18}{figure.2.4}%
\contentsline {figure}{\numberline {2.5}{\ignorespaces Max pooling (Stanford University, 2022).}}{19}{figure.2.5}%
\addvspace {10\p@ }
\contentsline {figure}{\numberline {3.1}{\ignorespaces The Kalobeyei and Dzaleka camps respective location in East Africa,}}{22}{figure.3.1}%
\contentsline {figure}{\numberline {3.2}{\ignorespaces The Kakuma-Kalobeyei land use and planning areas (UN-HABITAT, 2018)}}{24}{figure.3.2}%
\contentsline {figure}{\numberline {3.3}{\ignorespaces RGB UAV imagery of the Kalobeyei settlements in rural Turkana from OpenAerialMap}}{25}{figure.3.3}%
\contentsline {figure}{\numberline {3.4}{\ignorespaces The main Dzaleka Refugee Camp and the Katubza extension plan (Dzaleka North) designed by Urban Design Advisor to the UNHCR Werner Schnellenberg (Gross G., 2021).}}{26}{figure.3.4}%
\contentsline {figure}{\numberline {3.5}{\ignorespaces Digitised rooftop of the Dzaleka and Dzaleka North camps by HOT volunteers}}{27}{figure.3.5}%
\contentsline {figure}{\numberline {3.6}{\ignorespaces Motion artefacts unique to UAV imagery}}{28}{figure.3.6}%
\contentsline {figure}{\numberline {3.7}{\ignorespaces Perhaps geometric augmentation of horizontal flipping shall not be applied on the MNIST number of 5}}{31}{figure.3.7}%
\contentsline {figure}{\numberline {3.8}{\ignorespaces An example of Inverse RGB augmentation applied to the Train dataset.}}{32}{figure.3.8}%
\contentsline {figure}{\numberline {3.9}{\ignorespaces Collections of diverse and heterogeneous rooftops from the Kalobeyei, Dzaleka, and Dzaleka North datasets.}}{34}{figure.3.9}%
\contentsline {figure}{\numberline {3.10}{\ignorespaces The Encoder-Decoder U-Net architecture (Ronneberger et al., 2015, Seale et al., 2022)}}{35}{figure.3.10}%
\contentsline {figure}{\numberline {3.11}{\ignorespaces EfficientNet family Top 1\% Accuracy Assessment in ImageNet (Tan \& Le, 2020).}}{36}{figure.3.11}%
\contentsline {figure}{\numberline {3.12}{\ignorespaces The Confusion Matrix}}{40}{figure.3.12}%
\contentsline {figure}{\numberline {3.13}{\ignorespaces Examples of theoretical binary building classification.}}{41}{figure.3.13}%
\contentsline {figure}{\numberline {3.14}{\ignorespaces Schematic diagram of Intersection-over-Union}}{44}{figure.3.14}%
\contentsline {figure}{\numberline {3.15}{\ignorespaces Project workflow}}{45}{figure.3.15}%
\contentsline {figure}{\numberline {3.16}{\ignorespaces Simplified 5 steps project workflow with reference to \ref {fig:ETL_flow}}}{46}{figure.3.16}%
\addvspace {10\p@ }
\contentsline {figure}{\numberline {4.1}{\ignorespaces Sample of binary segmentation output of various combinations of tested architecture and experiment setup.}}{48}{figure.4.1}%
\contentsline {figure}{\numberline {4.2}{\ignorespaces Class-based Accuracy Assesment metrics for respective CNN architectures and experiment input dataset.}}{49}{figure.4.2}%
\contentsline {figure}{\numberline {4.3}{\ignorespaces Regression plot for \textit {Precision} and \textit {Recall} change in relation to architectural depth-wise change.}}{51}{figure.4.3}%
\contentsline {figure}{\numberline {4.4}{\ignorespaces Detailed strip plot for \textit {Precision} and \textit {Recall} change in relation to dataset input change.}}{53}{figure.4.4}%
\contentsline {figure}{\numberline {4.5}{\ignorespaces Regression plot for \textit {Precision} and \textit {Recall} change in relation to architectures' initalised weight change.}}{54}{figure.4.5}%
\contentsline {figure}{\numberline {4.6}{\ignorespaces Ambiguity which arise from labelling could cause a $True\ Positive$ prediction to be classified as $False\ Positive$.}}{56}{figure.4.6}%
\addvspace {10\p@ }
\addvspace {10\p@ }
\contentsline {figure}{\numberline {6.1}{\ignorespaces The algorithm of Adam (Kingma \& Ba., 2017).}}{74}{figure.6.1}%
\contentsline {figure}{\numberline {6.2}{\ignorespaces Compound scaling of the EfficientNet (Tan \& Le, 2020)}}{75}{figure.6.2}%