Hi Matthias, I'm glad to hear from you about the latest progress on the project. Congratulations on your completion of the main statistic analysis of the whole match, and thanks for pointing me to the Ascensio website and giving me the position data. About your interest on "semantic analysis" and "macro analysis", I'm not aware there exists some running system that can do the job automatically for the domain of soccer/football. I'm sorry I could not help on this aspect under your current time constraints. As regards your idea of visualizing human kinetics in selected match situations, I can try my best to do something for you. My program roughly approximates and visualizes the grand reaction force and major joint torques in simple human planar motions. It needs as input the (x, y) coordinates of major joints in each image frame and the height and body weight of the subject. My program (MATLAB scripts) provides an interface to collect those coordinates manually, which I used for my course project. Please provide me 1 to 3 video clips you are most interested in doing the visualization, together with the heights and body weights of the specific players of interest. I will try my program on one of them and send back to you the result avi video (which may last for only several seconds) to get some feedback. Feel free to contact me if you have any question. Cheers, Fahong On Wed, 21 Mar 2007, Matthias Rajmann wrote: > Date: Wed, 21 Mar 2007 19:23:32 +0100 > From: Matthias Rajmann > To: 'Fahong Li' > Subject: Harun Farocki, "Football" for documenta 2007, Final of FIFA WM 2006 > > HARUN FAROCKI FILMPRODUKTION > > MATTHIAS RAJMANN > > Gietlstr. 7 > > D - 81541 München / Munich > > Tel.: +49( 0)89 201 35 47 > > Fax.: +49( 0)721 15 13 78 521 > > Mobil:+49 (0)175 84 83 834 > > matthias.rajmann@farockifilm.de > > www.farockifilm.de > > > > > > Dear Fahong Li, > > > > please excuse my very late response to your interesting e-mail of Feb., 21.. > I had a look at your homepage and I find it very interestig how you seem to > be able to predict the action on the field. Also I like the kind of graphics > - the rather simple but very good readable visualization. > > > > Answering your question how to visualize all the analytical information I > want to let you know about the software Match Expert of Ascensio: > http://www.footballsoftpro.com/ > > > > Ascensio did work out the whole final in Match Expert. We are allowed to > visualize both in 2 D and 3 D of different angles and using several kinds of > graphs, f.e. relation between certain match players, speeds, trajectories, > heat maps, etc.. > > > > The modification of the software and especially the implementation of a > video realtime capturing module took all of my attention; that's why I reply > so late. Sorry again! > > > > You see that the main analytical part on more or less statistic base already > is done. > > > > What we are very much interested in is f.e. > > > > * semantic analysis - automated recognition with help of intelligent > systems of f.e. offending or defendig situations > > * macro analysis > > > > I read about you project and saw the video samples: Visualization of Human > Kinetics. > > > > We have the idea that it could be interesting to choose specific situations > of the match and to "feed" your software for Visualization of Human > Kinetics. > > > > The main questions are: > > > > * Is it possible to track out of given video (close ups of bodies, > panning/tilting camera) the articulations of a player? > > * Do you have the time to make tests and getting results in a rather > short period of time? > > > > We think that a cooperation would be feasable only when there is a running > system existing; too many modifications would mean lots of stress, doesn't > it? > > > > Please feel free for every kind of suggestion of what could be feasable! > > > > The exposé doesn't exist in an English version; but its content is the same > as the written explanation that had been forwarded of Hans-Dieter Burkhard. > > > > It would be great to cooperate with you and present your work in our > project! > > > > Here I give you access to the position data files - for that you can have a > look at it: > > > https://mc.1und1.de/mc/MO96USy0TbfFbhi60baVoToxQYMlZL > > > > > > Please don't hesitate to contact me. > > > > > > Best regards, > > > > Matthias > > > > -----Ursprüngliche Nachricht----- > Von: Fahong Li [mailto:fhli@cs.ubc.ca] > Gesendet: Mittwoch, 21. Februar 2007 04:37 > An: matthias.rajmann@farockifilm.de > Cc: Hans-Dieter Burkhard; Bob Woodham > Betreff: Harun Farocki, "Football" for documenta 2007, Final of FIFA WM > 2006T (fwd) > > > > Dear Matthias, > > > > I'm a graudate student at the University of British Columbia, Vancouver, > > Canada. I have been doing research in computer vision and artificial > > intelligence with sports game (ice hockey, soccer) analysis as the test > > domain. I learned your project "Football" for Documenta 12 from Dr. > > Burkhard through the robocup-sim mailing list. I think it is an > > interesting and exciting project; upon finished it will bring more public > > attention to recent research results in computer vision. So I'm willing to > > do something for your project. > > > > I did my M.Sc. thesis based on the IRIS TRA project, which describes and > > analyzes players' actions/trajectories in selected ice hockey game > > situations. Our proof-of-concept framework takes as input actual player > > motion trajectory data tracked from game video and supported by knowledge > > of hockey strategy, game context and specific player profiles. Currently > > It only outputs simple textual natural language descriptions and 2D > > graphic animations of the analysis. It could be extended to visualize the > > trajectories back into the images as long as the registrations between the > > image and the hockey rink are good enough. More details on my thesis > > project available on my home page: http://www.cs.ubc.ca/~fhli > > > > I think the programs developed in my thesis project might be adapted to > > your "Football" project according to my current understanding to the > > project's requirements listed in your previous message to Dr. Burkhard, > > i.e.: > > > > Quotation start > > ================ > > > > We hope to illuminate the layers and sequences of space, movement, > > intention and results and to make the individual elements visible by means > > of electronic image processing. The investigation will take place purely > > on the visual level, without spoken commentary. > > > > The possibilities we envisage include, > > > > * Presenting the match in such a way that the electronic image processing > > clearly marks those areas of the pitch in colour where play is not > > currently taking place. Two thirds of the roughly 50,000 m^2 making up a > > pitch often remain unused throughout a match. > > > > * Presenting a match so as to make the paths taken by one or more specific > > players visible on screen. > > > > * Presenting a match in a way that allows a visualisation of the fields of > > vision of one or more specific players in terms of cones shown on screen - > > when looking up from the ball, when checking their position, when looking > > round at the opposition. > > > > * A player has the ball and wants to pass. What are the other 9 players he > > could pass to doing at that moment? Are they marked, are they trying to > > break free? Are there any free spaces into which they could play the ball? > > > > ...... > > > > Our main materials are: > > > > * International Broadcast Feed - edited for TV > > * Clips Feed - Slow-Motion Replay > > * Tactical Feed - Birds eye view of the match > > * High Angle Feed - under stadium roof > > * Team Cam Italy - Bench shots etc > > * Team Cam France - Bench shots etc. > > * Player Cam Italy - Player close ups > > * Player Cam France - Player close ups > > > > Out of this feed we have gained the tracking data with positions of the > > players and the ball. > > > > ================ > > End of Quotation > > > > I wonder whether you have more specific requirements/emphases on how to > > visualize all the information above since the combined visualization > > results imposed on the images might be too complex to understand. In > > addition, what's the format (coordinate frame) of the position data? > > > > Please let me know if you are interested in trying our framework and/or > > if you have any further questions on our framework. And if possible, could > > you please provide an English version of the "documenta Fussball > > Expose.pdf" ( > > https://bugs.cs.ubc.ca/twiki/pub/Main/FahongLiDocumentaFootball/documentaFus > sballExpose.pdf > > ) I got from Dr. Burkhard? Thanks in advance. > > > > > > Yours Sincerely, > > Fahong > > > > ------------------------------------------------- > > Fahong LI 604-827-3986(lab phone) > > Graduate Student http://www.cs.ubc.ca/~fhli > > Department of Computer Science > > University of British Columbia > > 2366 Main Mall, Vancouver, BC, V6T 1Z4, Canada > > ------------------------------------------------- > > > > > > ---------- Forwarded message ---------- > > Date: Tue, 20 Feb 2007 13:55:55 +0100 (MET) > > From: Hans-Dieter Burkhard > > To: fhli@cs.ubc.ca > > Cc: hdb@informatik.hu-berlin.de > > Subject: Harun Farocki, "Football" for documenta 2007, Final of FIFA WM > 2006T > > > > Dear Fahong, > > > > as written in my last email to the community, I ask you > > to get in direct contact to > > > > matthias.rajmann@farockifilm.de > > > > Please discuss with him the options for a common work > > (and please let me know the results). > > > > I add further materials (sorry, only in German) in > > the attachment. > > > > Best regards > > Dieter > > > > ----------------------------------------- > > > >> From hdb@informatik.hu-berlin.de Tue Feb 20 06:24:02 2007 > > Date: Tue, 20 Feb 2007 13:45:37 +0100 (MET) > > From: Hans-Dieter Burkhard > > To: robocup-sim@cc.gatech.edu, sserver-three-d@lists.sourceforge.net, > > robocup-exec@cc.gatech.edu > > Subject: Re: [robocup-sim] Project: Annotations for FIFA soccer finale 2006 > > > > Dear all, > > > > after further discussions with Matthias (the producer for > > the Documenta project), the situation appears as follows: > > > > They need only one run of an appropriate tool to produce > > the annotation, comments etc. for the final FIFA match. > > (Of course, credits will be given for the tool.) > > At Documenta, only the results of this run (e.g. a video) > > need to be shown. > > > > The production has to be ready about end of April. > > It means that only existing tools are useful (only slight > > modifications), because there needs to be time for > > production. I dont see the possibility of a joint project > > (too much time needed). > > > > I got 3-4 proposals from the community. Forv these teams, I will > > make a direct contact to Matthias, such that you can discuss > > further details directly. > > > > Best regards > > Dieter > > > > robocup-sim mailing list > > robocup-sim@cc.gatech.edu > > https://mailman.cc.gatech.edu/mailman/listinfo/robocup-sim > > ---------------------------------------- > >