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 on
an unsupervised classification of the players based on clustering of their
trajectories which are extracted trajectories, automatically
extracted from
the 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 the
fabric 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 been
achieved 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 similarity
measure 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.