Live Soccer Analysis

In the billionaire industry of professional soccer, strategic planning is crucial to achieve the desire results. In this planning is often used information about players positions and trajectories which can be collected in by automatic video systems.

The Problem

Nowadays these systems are quite expensive due to their complex installation and an excessive human intervation in some semi-automatic procedures. Because of this, a lot of teams can't enjoy the advantages of automatic live game analysis.

Our Solution

In this project we pretend to study different architectures of a low-cost system for automatic video analysis on indoor soccer games based in the utilization of small unmanned air vehicles such as Quadcopters.


Soccer is one of the most popular sports in the world being actually dubbed as the king of the sports. The industry behind this sport justifies the millionaire budgets of the teams to achieve the best possible results and, of course, the respective financial return. This budget is used not only to purchase players but also in technical staff responsible for technical and tactical planning of the games. Data about the position and trajectories of the players and the ball are essential for the staff to understand and correct some usual behaviors of the team and opponents allowing the coach to have the best strategy during a game. Nowadays, the automatic vision systems found in the market are quite expensive due to their hard installation and excessive human intervention. In this project we pretend to study a low cost automatic vision system for indoor soccer analysis of individual and collective performance and also other game events. This system will be based on the utilization of unmanned air vehicles as the Quadcopter Ar.Drone 2.0 with a High Definition camera onboard.

Main Goals

In this project is pretended to study the feasibility of a system based on one or more Quadcopter to capture images from an indoor soccer game. From these images we want to extract useful information about individual and team performance on an automatic way with minimal human intervention. On a first stage, we will need to design a flight control to the quadcopter in order to capture the images of the game action in an automatic and not intrusive way using the frontal camera of the vehicle. Remotly, the collected images will be processed to calcule controll commands to the Quadcopter. The objective is to capture relevant game action and simultaneously to not interact improperly with any element of the game (ball, players, assistance, etc). From the video sequences colected information about ball and players position and trajectories will be extracted. In the end all the information and data will be presented in a clear way on an user-friendly interface.