\newcommand{\tmptw}{.8}
\begin{comment}
python -m ibeis.scripts.specialdraw merge_viewpoint_graph --dpath ~/latex/cand/ --save figures4/bridgethegap2.png --figsize=15,15 --clipwhite --diskshow
python -m ibeis.scripts.specialdraw merge_viewpoint_graph --dpath ~/latex/cand/ --save figures4/bridgethegap2.png --figsize=15,15 --clipwhite --diskshow --constraint
\end{comment}
\SingleImageCommand{bridgethegap}{}{A merge case between distinct viewpoints}{
% ---
    A new viewpoint matches a back-left and a back-right viewpoint and
      merges two names.
    The matches that connect the left and right viewpoint are shown in
      blue.
    %is able to  two disjoint sets of annotations.  Two disjoint viewpoints are
    %merged by a common transition viewpoint.
% ---
}{figures4/bridgethegap2.png}



\begin{comment}
python -m ibeis.annotmatch_funcs review_tagged_splits --zoom= --dpath ~/latex/cand/ --save figures4/splitcase.png --figsize=15,15 --clipwhite --diskshow
python -m ibeis.annotmatch_funcs review_tagged_splits --dpath ~/latex/cand/ --save figures4/splitcase.jpg --figsize=10,4 --dpi 180 --clipwhite --diskshow --zoom=.7

aids = [14397, 15120, 15600, 15606, 16804, 16930, 16967, 17012, 17062]

\end{comment}
\SingleImageCommand{splitcase}{\tmptw}{
    A split case
}{
% ---
A graph of matches showing a split case.
Two pairs of annotations have been incorrectly labeled with the same name.
The two edges marked in orange have been flagged as a potential split case.
This split case can be resolved by cutting the highlighted edges.
% ---
}{figures4/splitcase.png}




\begin{comment}
python -m ibeis.annotmatch_funcs review_tagged_joins --dpath ~/latex/cand/ --save figures4/mergecase.png --figsize=15,15 --clipwhite --diskshow
python -m ibeis.annotmatch_funcs review_tagged_joins --zoom=.5 --dpath ~/latex/cand/ --save figures4/mergecase.jpg --figsize=10,4 --dpi 260 --clipwhite --diskshow
\end{comment}
\SingleImageCommand{mergecase}{\tmptw}{A merge case}{
% ---
An example of a merge case found in the \GZC{}.
Two sets of annotations are incorrectly marked as different individuals.
The two annotations marked in orange have been flagged as a potential merge
  case.
This merge case can be resolved by drawing an edge between the highlighted
  annotations.
% ---
}{figures4/mergecase.png}


\begin{comment}
python -m ibeis.algo.preproc.preproc_occurrence compute_occurrence_groups --dpath ~/latex/cand/ --save figures4/occurgraph.png --figsize=40,40 --clipwhite --diskshow
python -m ibeis.algo.preproc.preproc_occurrence compute_occurrence_groups --dpath ~/latex/cand/ --save figures4/occurgraph.png --figsize=7,7 --dpi 220 --clipwhite --diskshow
python -m ibeis.algo.preproc.preproc_occurrence compute_occurrence_groups --show --zoom=.8 --dpath ~/latex/cand/ --save figures4/occurgraph.jpg --figsize=30,10 --dpi 160 --clipwhite --diskshow
\end{comment}
%\SingleImageCommand{occurgraph}{.8\tmptw}{occurgraph}{
%occurgraph.jpg
%}{figures4/occurgraph.jpg}

\begin{comment}
python -m ibeis.viz.viz_image gcshow_multi_images --db PZ_Master1 --gids=22825,22186,22187,22828,22840,22842,22843,22847,22848,22849,22852,22121,22122,22124,22125,22126,22642,22643,22644,22645,22646,22647,22826 --adjust=.05 --dpath ~/latex/cand/ --save "figures4/occurimgs.jpg" --figsize=9,4 --clipwhite --dpi=180 --diskshow


python -m ibeis.viz.viz_image gcshow_multi_images --db PZ_Master1 --gids=22825,22186,22187,22828,22840,22842,22843,22847,22848,22849,22852,22121,22122,22124,22125,22126,22642,22643,22644,22645,22646,22647,22826 --adjust=.05 --dpath ~/t2 --save "occurimg.jpg" --figsize=18,8 --clipwhite --dpi=280 --diskshow --saveparts

python -m ibeis.viz.viz_image gcshow_multi_images --db PZ_Master1 --gids=1,100,200,300,400,500,600,700,800,900,1000,1100,1200,1300,1400,1500,1600,1700,1800,1900,2000 --adjust=.05 --dpath ~/t2 --save "notoccurimg.jpg" --figsize=18,8 --clipwhite --dpi=280 --diskshow --saveparts

python -m ibeis.viz.viz_image gcshow_multi_images --db PZ_Master1 --gids=22642,22643,22644 --adjust=.05 --dpath ~/t2 --save "encimg.jpg" --figsize=18,8 --clipwhite --dpi=280 --diskshow --saveparts

python -m ibeis.viz.viz_image gcshow_multi_images --db PZ_Master1 --gids=254,14911,15050,15247,15377,15405,15535,7530,4138,7713,9807,9934,24279,24579,25495,25496,21719,23602,21019,21022,22825,22826,22828,22121,22122,22124,22125,22186,22126,22187,22840,22642,22643,22842,22843,22644,22847,22848,22645,22849,22646,22647,22852 --adjust=.05 --dpath ~/t3 --save "occurimg2.jpg" --figsize=18,8 --clipwhite --dpi=280 --diskshow 


python -m ibeis.viz.viz_image gcshow_multi_images --db PZ_Master1 --gids=1,100,200,300,400,500,600,700,800,900,1000,1100,1200,1300,1400,1500,1600,1700,1800,1900,2000 --adjust=.05 --dpath ~/t2 --save "occurimg.jpg" --figsize=18,8 --clipwhite --dpi=280 --diskshow --saveparts
\end{comment}

\MultiImageCommandII{intraoccur}{.4}{
    \Intraoccurrence{} matching}{
% ---
\Intraoccurrence{} matching.
\Cref{sub:intraoccurA} shows a group of images in an \occurrence{} with
  bounding boxes around unknown annotations.
\Cref{sub:intraoccurB} illustrates the result of \intraoccurrence{}
  matching, which is a graph where an edge between two annotations
  denotes that pair is a correct match.
The connected components (denoted by color) of this graph are the
  \encounters{}.
% ---
}{figures4/occurimgs.jpg}{figures4/occurgraph.png}


\begin{comment}
python -m ibeis.scripts.specialdraw intraoccurrence_connected --dpath ~/latex/cand/ --save figures4/precut.png --figsize=40,40 --clipwhite --diskshow
\end{comment}

\begin{comment}
python -m ibeis.scripts.specialdraw intraoccurrence_connected --dpath ~/latex/cand/ --save figures4/postcut.png --figsize=40,40 --clipwhite --diskshow --postcut
\end{comment}
\MultiImageCommandII{gencutfig}{.44}{
    Identification graphs
}{
% ---
The identification algorithm builds a graph of potential matches.
\Cref{sub:gencutfigA} shows the input graph to the identification
  algorithm.
Blue edges are drawn between the \uset{}, orange edges are drawn
  between the \uset{} and \lset{}, and colored edges corresponding to
  \names{} are drawn between nodes in the \lset{}.
A colored border is drawn around annotations in the \lset{}.
\Cref{sub:gencutfigB} shows the result of identification, which is a
  graph where an edge between two annotations denotes that
  pair correctly matches.
The connected components denoted by color of this graph are the new
  individuals.
% ---
}{figures4/precut.png}{figures4/postcut.png}


\begin{comment}
python << endpython
def test():
    return bar
endpython
pip install git+https://github.com/johnyf/nx2tikz.git
\end{comment}


%For Masai giraffes
%$\nvisit_1 = 89$
%$\nvisit_2 = 17$
%$\resight = 10 $
%%index = 151.3
%%variance = 60.175479378231799


\begin{comment}
python -m utool.util_latex gcmake_score_tabular:0 --show
python << endpython
import sympy as sym
nvisit_1, nvisit_2, resight = sym.symbols('c_1 c_2 m')
sqrt = sym.sqrt
c1 = nvisit_1 = 940
c2 = nvisit_2 = 433
c3 = resight  = 115
sqrt = np.sqrt
index = (nvisit_1 * nvisit_2) / resight 
variance = 1.96 * sqrt((nvisit_1 ** 2 * nvisit_2 * (nvisit_2 - resight)) / (resight ** 3))
print('index = %r' % (index,))
print('variance = %r' % (variance,))
import utool as ut
row_lbls = ['Plains zebras', 'Masai giraffes']
col_lbls = [r'$\nvisit_1$', r'$\nvisit_2$', r'$\resight$', r'$\poptotal$']
values = [['$940$', '$422$', '$114$', '$3539.3 \pm 554.3$'], 
          ['$89$',   '$17$',  '$10$', '$151.3 \pm 60.17$']]

title='Estimated populations of plains zebras and Masai giraffes in Nairobi National Park from the \\GZC{}.'
label = 'popest'

tabularstr = ut.make_table2(row_lbls, col_lbls, values, title=title, label=label)
tabularcmd = ut.latex_newcommand(ut.latex_sanatize_command_name(label), tabularstr)
print(tabularcmd)
endpython
\end{comment}


\newcommand{\popest}{
\begin{table}[ht!]
\centering
%\begin{tabular}{|c|rrr|c|}
\caption[\caplbl{tbl:popest}Results of the \GZC{}]{
% ---
\caplbl{tbl:popest}Results of the \GZC{}.
For each species, the identification algorithm was used to determine:
($\nvisit_1$) the number of individuals sighted on March 1\st{},
  ($\nvisit_2$) the number of individuals sighted on March 2\nd{},
  ($\resight$) and the total number of resighted individuals.
The Lincoln-Petersen index (\Cref{eqn:lpi}) was then used to estimate
  ($\poptotal$) the total population size of plains zebras and Masai
  giraffes in Nairobi National Park.
% ---
}
\label{tbl:popest}
\begin{tabular}{l rrr c}
\toprule
Species        & $\nvisit_1$ & $\nvisit_2$ & $\resight$ & $\poptotal$       \\
\midrule
Plains zebras  & $940$       & $422$       & $114$      & $\mathbf{3539.3} \pm 554.3$\\
\midrule
Masai giraffes & $89$        & $17$        & $10$       & $\mathbf{151.3} \pm 60.17$ \\
\bottomrule
\end{tabular}
\end{table}
}



\begin{comment}
        python -m ibeis.algo.hots.testem test_em --diskshow --no-cnn --dpath ~/latex/cand/ --save "figures4/emgraphprior.png" --figsize=11,4 --clipwhite --dpi=180 --diskshow
        python -m ibeis.algo.hots.testem test_em --diskshow --no-cnn --dpath ~/latex/cand/ --save "figures4/emgraphpost.png" --postem --figsize=11,4 --clipwhite --dpi=180 --diskshow
\end{comment}
\MultiImageCommandII{emgraph}{.4}{An EM-based identification graph}{
    % ---
    The graph associated with the expectation maximization algorithm.
    The red $\nid$ nodes represent potential labels for the blue $\X$
      annotation nodes.
    The measure pairwise probability of being the same is shown on
      edges between annotations.
    The probabilities of name labelings are shown on the orange edges.
    The prior probabilities of name labelings are shown on the left.
    The posterior probabilities of the name labelings are shown on the
      right.
    Note that we chose to constrain the number of labels to be
      different from the number of annotations, but this need not be the
      case.
    % ---
}{figures4/emgraphprior.png}{figures4/emgraphpost.png}


\begin{comment}
    python -m ibeis.algo.hots.bayes make_name_model --num-annots=2 --dpath ~/latex/cand/ --save "figures4/bayesII.jpg" --figsize=9,4 --clipwhite --dpi=180 --diskshow
    python -m ibeis.algo.hots.bayes make_name_model --num-annots=3 --dpath ~/latex/cand/ --save "figures4/bayesIII.jpg" --figsize=9,4 --clipwhite --dpi=180 --diskshow
    python -m ibeis.algo.hots.bayes make_name_model --num-annots=4 --dpath ~/latex/cand/ --save "figures4/bayesIV.jpg" --figsize=9,4 --clipwhite --dpi=180 --diskshow
\end{comment}
%\MultiImageCommand{bayesnets}{.3}{
%    % ---
%    The proposed Bayesian network with 2, 3, and 4 annotations.
%    Note that the variables used here are different than in other
%      sections.
%    The top row contains a node for each anotations representing its
%      probability distribution over the possible name labelings.
%    The middle row represents if two annotations are the same or not.
%    The bottom row holds the observed probabilities of two annotations
%      being the same.
%    % ---
%}{figures4/bayesII.jpg}{figures4/bayesIII.jpg}{figures4/bayesIV.jpg}
\newcommand{\bayesnets}{
\begin{figure}[ht!]
\centering
\begin{subfigure}[h]{0.15\textwidth}
\centering
%\includegraphics[width=\textwidth]{figures1/ThreeSixtyFigureA.jpg}\caption{}\label{sub:ThreeSixtyFigureA}
\includegraphics[width=\textwidth]{figures4/bayesII.jpg}\caption{}\label{sub:bayesnesA}
\end{subfigure}
~~% --
\begin{subfigure}[h]{0.24\textwidth}
\centering
\includegraphics[width=\textwidth]{figures4/bayesIII.jpg}\caption{}\label{sub:bayesnesB}
\end{subfigure}
~~% --
\begin{subfigure}[h]{0.47\textwidth}
\centering
\includegraphics[width=\textwidth]{figures4/bayesIV.jpg}\caption{}\label{sub:bayesnesC}
\end{subfigure}
\caption[\caplbl{bayesnets}The identification graph as a Bayses net]{\caplbl{bayesnets}
% ---
    The identification graph in the form of a Bayesian network with 2, 3, and 4 annotations.
    Note that the variables used here are different than in other sections.
    The top row contains a node for each anotations representing its probability distribution over the possible name labelings.
    The middle row represents if two annotations are the same or not.
    The bottom row holds the observed probabilities of two annotations being the same.
% ---
}
\label{fig:bayesnets}
\end{figure}
}

\begin{comment}
    python -m ibeis.algo.hots.bayes make_name_model --num-annots=2 --modeltype=markov --dpath ~/latex/cand/ --save "figures4/markovII.jpg" --figsize=18,8 --clipwhite --dpi=180 --diskshow
    python -m ibeis.algo.hots.bayes make_name_model --num-annots=3 --modeltype=markov --dpath ~/latex/cand/ --save "figures4/markovIII.jpg" --figsize=18,8 --clipwhite --dpi=180 --diskshow
    python -m ibeis.algo.hots.bayes make_name_model --num-annots=4 --modeltype=markov --dpath ~/latex/cand/ --save "figures4/markovIV.jpg" --figsize=18,8 --clipwhite --dpi=180 --diskshow
\end{comment}
\MultiImageCommand{markovnets}{.3}{
    The corresponding markov networks for the proposed Baysian models.
}{figures4/markovII.jpg}{figures4/markovIII.jpg}{figures4/markovIV.jpg}

\begin{comment}
    python -m ibeis.algo.hots.bayes make_name_model --num-annots=2 --modeltype=junc --dpath ~/latex/cand/ --save "figures4/juncII.jpg" --figsize=18,8 --clipwhite --dpi=180 --diskshow
    python -m ibeis.algo.hots.bayes make_name_model --num-annots=3 --modeltype=junc --dpath ~/latex/cand/ --save "figures4/juncIII.jpg" --figsize=18,8 --clipwhite --dpi=180 --diskshow
    python -m ibeis.algo.hots.bayes make_name_model --num-annots=4 --modeltype=junc --dpath ~/latex/cand/ --save "figures4/juncIV.jpg" --figsize=18,8 --clipwhite --dpi=180 --diskshow
\end{comment}
\MultiImageCommand{juncnets}{.3}{
    The corresponding junction trees corresponding to the cliques in the markov
    network. These structures are used for belief propogation on the proposed
    Baysian models.
}{figures4/juncII.jpg}{figures4/juncIII.jpg}{figures4/juncIV.jpg}


\begin{comment}
python -m ibeis.scripts.specialdraw --exec-merge_viewpoint_graph \
    --dpath ~/latex/cand/ \
    --save figures4/bridgethegap.png --figsize=8,8 --dpi 220 \
    --clipwhite --diskshow
\end{comment}
\SingleImageCommand{Rhombicuboctahedron}{.3 }{The rhombicuboctahedron}{
% ---
    We propose to model viewpoints and viewpoint differences using
    the rhombicuboctahedron.
% ---
}{figures4/Rhombicuboctahedron.jpg}


\begin{comment}
    python -m ibeis.scripts.specialdraw general_identify_flow --save figures4/pairprob.png --diskshow --clipwhite --dpath ~/latex/cand/ --figsize=24,10
\end{comment}
\SingleImageCommand{pairprob}{1}{Annotation-vs-annotation probability flowchart}{
% ---
    Information flow of the annotation-vs-annotation similarity.
    Local and global features are combined into a constant length
      ``similarity vector''.
    A classifier learns to label a similarity vector as either matching or
      not matching.
    A probability is extracted from the output of the of this classifier.
    This probability will be used later in the in identification algorithm
      as edge weights on a graph.
% ---
}{figures4/pairprob.png}

\begin{comment}
    python -m ibeis.scripts.specialdraw graphcut_flow --save figures4/cutiden.png --diskshow --clipwhite --dpath ~/latex/cand/ --figsize=24,10
\end{comment}
\SingleImageCommand{cutiden}{1}{Graph-based identification flowchart}{
% ---
    Given the algorithm to produce a match probability between two
    annotations the identification algorithm first generates a graph of
    potential matches and then cuts edges resulting in identifications.
% ---
}{figures4/cutiden.png}


\begin{comment}
    python -m ibeis.scripts.specialdraw intraoccurrence_connected --show --small --save figures4/edgeexample.png --diskshow --clipwhite --dpath ~/latex/cand/ --figsize=24,10
\end{comment}
\SingleImageCommand{edgeexample}{1}{Edge types for graph cuts}{
% ---
    The different types of edges in the graph.
    The \lset{} annotations have colored borders denoting a known
      \name{} label.
    The \uset{} annotations have colored borders denoting an unknown
      \name{} label.
    Each potential match is represented with an edge.
    Orange edges are between the \uset{} and \lset{}.
    Purple edges are within the \uset{}.
    Other colored edges are between the \lset{}.
% ---
}{figures4/edgeexample.png}


\begin{comment}
    python -m ibeis.scripts.specialdraw intraoccurrence_connected --show --smaller --save figures4/graphexample.png --diskshow --clipwhite --dpath ~/latex/cand/ --figsize=24,10
\end{comment}
\SingleImageCommand{graphexample}{1}{An identification graph}{
% ---
    A simple identification graph.
    Edges are weighted using a learned probability that the annotations are
      matching, not matching, or not comparable.
    The goal is to partition this graph into by individual identity.
    The colors in this graph denote different edge types which are discussed
      in~\cref{subsec:graphinf}.
% ---
}{figures4/graphexample.png}


\begin{comment}
    python -m ibeis.core_annots --test-compute_one_vs_one --aids=5245,5161 --show --db PZ_Master1 --save figures4/leftrightfacematch.png --diskshow --clipwhite --dpath ~/latex/cand/ --figsize=24,10 

      % PZ_Master1 aids=5161,5245
      %python -m ibeis --tf make_name_graph_interaction --db PZ_Master1 --aids=5161,5245 --show --no-with-all
      % python -m ibeis.core_annots --test-compute_one_vs_one --aids=5161,5245 --show --db PZ_Master1
      % Fix the one-vs-one algorithm to show this
\end{comment}
\SingleImageCommand{leftrightfacematch}{\tmptw}{Matchable annotations with incompatible viewpoints}{
    % ---
    The above annotations are able to be matched even though they have
      been assigned incompatible \vpFrontLeft{} and \vpFrontRight{}
      viewpoints.
    This is because we assign viewpoint as a global attribute of an
      annotation instead of labeling viewpoints of each animal's parts.
    % ---
}{figures4/leftrightfacematch.png}

\begin{comment}
    python -m ibeis.scripts.specialdraw setcover_example --show --save figures4/exemplarcover.png --diskshow --clipwhite --dpath ~/latex/cand/ --figsize=24,10  --arrow-width=.4 --line-width=1
\end{comment}
\SingleImageCommand{exemplarcover}{\tmptw}{Example of exemplar selection}{
% ---
    An example of exemplar selections.
    The number of requested exemplars is $3$.
    A directed arrow $\X \rightarrow \Y$ denotes that annotation $\Y$ is in
      the candidate covering set of annotation $\X$.
    A blue border is drawn around annotations chosen as exemplars.
% ---
}{figures4/exemplarcover.png}