Fixed merge conflict.

docs/report.tex

@@ -350,63 +350,71 @@ it as good as possible because all occurrences are in the learning set.
 
 To be able to use the code efficiently, we wrote a number of scripts. This
 section describes the purpose and usage of each script. For each script it is
-essential that you use the correct folder and subfolder naming scheme. The scheme
-is as follows:
+essential that you use the correct folder and subfolder naming scheme. The
+scheme is as follows:
 
 \begin{enumerate}
-    \item A main folder called `images' placed in the current directory as the
-    src folder.
+    \item A main folder called `images' placed in the root directory.
     \item In the images folder there have to be three folders.  Images, Infos
-    and LearningSet.
-    \item The Images and Infos folder contain subfolders which are numbered
+    characters
+    \item The Images and Infos folder contain subdirectories which are numbered
     ($0001$ to possibly $9999$).
-    \item In each of the subfolders the data (i.e the images or xml files) can
-    be placed.  And have to be named $00991_XXXXX.ext$, where XXXXX can be
+    \item In each of the subdirectories the data (i.e the images or xml files)
+    can be placed. And have to be named $00991_XXXXX.ext$, where XXXXX can be
     $00000 to 99999$.
-    \item For-loops in the script currently only go up to 9 subfolders, with a
-    maximum of containing 100 images or xml files. These numbers have to be
-    adjusted if the scripts are being used, but with a bigger dataset.
+    \item For-loops in the script currently only go up to 9 subdirectories,
+    with a maximum of containing 100 images or xml files. These numbers have to
+    be adjusted if the scripts are being used, but with a bigger dataset.
 \end{enumerate}
 
-It is of course possible to use your own naming scheme. A search for the
-$filename$ variable will most likely find the occurences where the naming
-scheme is implemented.
-
-
 \subsection*{\texttt{create\_characters.py}}
 
-
+Generates a file containing character objects with their feature vectors. Also,
+the learning set and test set files are created for the given combination of
+NEIGHBOURS and BLUR\_SCALE.
 
 \subsection*{\texttt{create\_classifier.py}}
 
-
+Generates a file containing a classifier object for the given combination of
+NEIGHBOURS and BLUR\_SCALE. The script uses functions from
+\texttt{create\_characters.py} to ensure that the required character files
+exist first. Therefore, \texttt{create\_characters.py} does not need to
+executed manually first.
 
 \subsection*{\texttt{find\_svm\_params.py}}
 
+Performs a grid-search to find the optimal value for \texttt{c} and
+\texttt{gamma}, for the given combination of NEIGHBOURS and BLUR\_SCALE. The
+optimal classifier is saved in
+\emph{data/classifier\_\{BLUR\_SCALE\}\_\{NEIGBOURS\}.dat}, and the accuracy
+scores are saved in in
+\emph{results/results\_\{BLUR\_SCALE\}\_\{NEIGBOURS\}.txt}.
+
+Like \texttt{create\_classifier.py}, the script ensures that the required
+character object files exist first.
+
+\subsection*{\texttt{run\_classifier.py}}
 
+Runs the classifier that has been saved in
+\emph{data/classifier\_\{BLUR\_SCALE\}\_\{NEIGBOURS\}.dat}. If the classifier
+file does not exist yet, a C and GAMMA can be specified so that it is created.
+Therefore, it is not necessary to run \texttt{create\_classifier.py} first.
 
 \subsection*{\texttt{generate\_learning\_set.py}}
 
 Usage of this script could be minimal, since you only need to extract the
 letters carefully and successfully once. Then other scripts in this list can
 use the extracted images. Most likely the other scripts will use caching to
-speed up the system too. But in short, the script will create images of a
-single character based on a given dataset of license plate images and
-corresponding xml files. If the xml files give correct locations of the
-characters they can be extracted. The workhorse of this script is $plate =
-xml_to_LicensePlate(filename, save_character=1)$. Where
+speed up the system too. But in short, the script will create images of a single
+character based on a given dataset of license plate images and corresponding
+XML files. If the XML files give correct locations of the characters they can
+be extracted. The workhorse of this script is \texttt{plate =
+xml\_to\_LicensePlate(filename, save\_character=1)}. Where
 \texttt{save\_character} is an optional variable. If set it will save the image
-in the LearningSet folder and pick the correct subfolder based on the character
-value. So if the XML says a character is an 'A' it will be placed in the 'A'
-folder. These folders will be created automatically if they do not exist yet.
-
-\subsection*{\texttt{load\_learning\_set.py}}
-
-
-
-\subsection*{\texttt{run\_classifier.py}}
-
-
+in the characters folder and pick the correct subdirectory based on the
+character value. So if the XML says a character is an 'A' it will be placed in
+the `A' folder. These folders will be created automatically if they do not
+exist yet.
 
 \section{Finding parameters}