deletions | additions       diff --git a/DeclareMathOperator_argmin_arg_min_DeclareMathOperator__1.tex b/DeclareMathOperator_argmin_arg_min_DeclareMathOperator__1.tex  index 0a1df1a..7880302 100644  --- a/DeclareMathOperator_argmin_arg_min_DeclareMathOperator__1.tex  +++ b/DeclareMathOperator_argmin_arg_min_DeclareMathOperator__1.tex 
...
In all cases, the result will be a labeled dataset of images and corresponding $x,y$-positions. The $x,y$-positions are of different quality depending on the used technique: orthomap-based, poster-based, or motion tracking-based.   \section{Pillar II: Texton-based Approach}  \label{sec:textons}  In this section, the core of the proposed algorithm, the  implementation of the proposed texton framework is described. The  histograms of textons are used as features for the $k$-Nearest  Neighbors ($k$NN) algorithm. The outputs of this regression technique  are possible $x,y$-coordinates for a given image.  In the following pseudo code, $M$ represents the number of particles,  $z_t^x$ the output of the texton framework at time $t$, and $f_t^x$  the estimated flow at time $t$.  Importantly, for generating the training dataset, no subsampling was  used. Therefore, the training dataset will not show any variance, if  the same images were used.