deletions | additions       diff --git a/section_Introduction_State_of_the__.tex b/section_Introduction_State_of_the__.tex  index bc955dd..519a2f6 100644  --- a/section_Introduction_State_of_the__.tex  +++ b/section_Introduction_State_of_the__.tex 
...
\section{Introduction}  State-of-the art works methods  on soccer match analysis use extra high level data and annotation annotations  (such as whoscored, transfertmark) which require human annotation and so preprocessing. In this project we limited our input to the video of a match filmed match, recorded  with a multi-camera system. We focus onan  unsupervised classification of the players based on clustering of their trajectories which are extracted trajectories,  automatically extracted  fromthe video of a  match video sequences  during a short period of time. Bases Based  on this classification information,  we want to be able to determine the leaders are interested on characterizing several aspects  of the game, match (such as detecting the game leaders,  to anticipate the displacement of the players or to find pattern patterns  in the paths of the players. This information is deeply woven into player paths). In such a way, we aim at obtaining details about  thefabric of  team strategy through the  analysis and thus lead to of  high level semantic data about the match. cues.  The problem of tracking players with multiple camera have cameras has  beenachieved  consistently studied,  for instance example  in \cite{Ben_Shitrit_2011} even if their paths may intersect over long period periods  of time. On another hand In this studies,  the clustering process is closely linked to the choice of a metric between explaining  the object similarities among the objects  we want to classify: here in our case,  the players' player  trajectories. Most of the time the metrics used are based on euclidian metric. However distance (YOU SHOULD CITE A REFERENCE HERE). However,  only discrete trajectories are available in the form of array which can be of different sizes sizes,  with different time discretization or with different speed and euclidian metric are irrelevant to measure trajectories interdependencies. speed. Euclidian distance would not result in a good choice for these cases.  So, in our case, a relevant clustering process should be decomposed into 2 parts: \begin{itemize}  \item Compute the distances matrix distance  between the trajectories we want to clusterize, producing a distance matrix.  \item Use a general clustering algorithm which is independent of the metric used metric, to produce the clusters.  \end{itemize}  Hence In this work,  we used the clustering algorithm described in \cite{NIPS2008_3478} which take a distance matrix as single input. Furthermore it It  automatically selects the clusters' cluster  centers and the number of clusters. Our contribution consist of a relevant novel metric, which can explain the  similaritymeasure  between the  trajectories based on Mutual Information which Information. It  allows the clustering of the players' player  trajectories according to the interdependency of their path.