The science of 'he shoots, he scores!'
Nov 11, 2003 0:52:51 GMT -5
Post by MC Habber on Nov 11, 2003 0:52:51 GMT -5
I thought this was interesting, at least from the perspective of a computer geek 
. One of my profs is involved in this.
The science of 'he shoots, he scores!'
Computer study tracks winning hockey manoeuvres
Doug Ward
Vancouver Sun
Monday, November 10, 2003
Before Kenji Okuma left his native Japan and began graduate studies in computer science at the University of B.C., he'd never watched hockey on television or laced up a pair of skates.
But over the past year-and-a- half, Okuma, 25, has watched more hockey broadcasts on Hockey Night in Canada or TSN than most die-hard Canucks' fans. He's logged hundreds of hours of games on videotape, paying special attention to goal highlights.
Okuma is part of a team of UBC computer scientists that is taking hockey analysis beyond Don Cherry into another dimension.
Okuma and four other researchers are developing a computer system capable of tracking the movements, or motion trajectories, of players in the National Hockey League.
The project is called the Acquisition, Querying and Prediction of Motion Trajectories. Its goal is to create an extensive database of NHL players in motion.
The idea is that the movement patterns of various players can be identified, allowing coaches to understand how players react and why some plays work and others don't.
The Japanese graduate student has developed the software that isolates and plots each player's route on the ice.
"It is very difficult for humans to track more than three objects at a time. But with computers you can track all the players on the scene simultaneously."
Okuma's software takes the videotaped images, which capture only a side or end view of the rink, and converts the broadcast footage into a digitized top-down view. This overhead view of players' movements is more useful for coaches or analysts.
"Once you get the data, you can analyze the movements and make predictions about future patterns," Okuma said. "You can take a goal scorer and see how he moves five seconds before each goal through a number of seasons. You can see if there is a pattern.
"You can see that if the players move one way, other players react a certain way."
Okuma said hockey was chosen for the project because the sport is so popular in B.C. Also, computerized movement tracking systems already exist for many other major sports, including soccer, baseball, football and tennis.
Okuma said he enjoys hockey but is growing of weary of watching game after game on television.
"Now, whenever I see the game, it reminds me of my research. So it's hard for me to relax and just enjoy the game."
Okuma, who completed his masters degree in computer science earlier this year at UBC, has been working on the project at the Lab for Computational Intelligence with his thesis advisers, Jim Little and David Lowe from the laboratory of computational intelligence at UBC.
They are part of a UBC project team, which also includes fellow computer scientists Robert Woodham and Raymond Ng.
The project is funded by the Institute for Robotics and Intelligent Systems (IRIS), a group that fosters collaboration among top Canadian researchers.
Okuma said his project team would like to get access to more camera angles at NHL rinks than are currently available on hockey broadcasts. One idea is to retrieve footage from the many private video cameras that are located at most NHL rinks.
The insights provided by the motion database aren't restricted to hockey coaches and analysts. Sports broadcasters could use it to provide a more elaborate play commentary.
The system could also help computer game and film industries create more realistic graphics, Okuma said. Eventually, the research could be used to develop robots capable of imitating human motion.
Okuma returns in December to Japan, where he intends to pursue a PhD and go back to being a fan of his favourite sport -- soccer.
www.canada.com/vancouver/news/story.asp?id=F3A4CD63-4CD3-454B-885A-EA607CE026F5
. One of my profs is involved in this.The science of 'he shoots, he scores!'
Computer study tracks winning hockey manoeuvres
Doug Ward
Vancouver Sun
Monday, November 10, 2003
Before Kenji Okuma left his native Japan and began graduate studies in computer science at the University of B.C., he'd never watched hockey on television or laced up a pair of skates.
But over the past year-and-a- half, Okuma, 25, has watched more hockey broadcasts on Hockey Night in Canada or TSN than most die-hard Canucks' fans. He's logged hundreds of hours of games on videotape, paying special attention to goal highlights.
Okuma is part of a team of UBC computer scientists that is taking hockey analysis beyond Don Cherry into another dimension.
Okuma and four other researchers are developing a computer system capable of tracking the movements, or motion trajectories, of players in the National Hockey League.
The project is called the Acquisition, Querying and Prediction of Motion Trajectories. Its goal is to create an extensive database of NHL players in motion.
The idea is that the movement patterns of various players can be identified, allowing coaches to understand how players react and why some plays work and others don't.
The Japanese graduate student has developed the software that isolates and plots each player's route on the ice.
"It is very difficult for humans to track more than three objects at a time. But with computers you can track all the players on the scene simultaneously."
Okuma's software takes the videotaped images, which capture only a side or end view of the rink, and converts the broadcast footage into a digitized top-down view. This overhead view of players' movements is more useful for coaches or analysts.
"Once you get the data, you can analyze the movements and make predictions about future patterns," Okuma said. "You can take a goal scorer and see how he moves five seconds before each goal through a number of seasons. You can see if there is a pattern.
"You can see that if the players move one way, other players react a certain way."
Okuma said hockey was chosen for the project because the sport is so popular in B.C. Also, computerized movement tracking systems already exist for many other major sports, including soccer, baseball, football and tennis.
Okuma said he enjoys hockey but is growing of weary of watching game after game on television.
"Now, whenever I see the game, it reminds me of my research. So it's hard for me to relax and just enjoy the game."
Okuma, who completed his masters degree in computer science earlier this year at UBC, has been working on the project at the Lab for Computational Intelligence with his thesis advisers, Jim Little and David Lowe from the laboratory of computational intelligence at UBC.
They are part of a UBC project team, which also includes fellow computer scientists Robert Woodham and Raymond Ng.
The project is funded by the Institute for Robotics and Intelligent Systems (IRIS), a group that fosters collaboration among top Canadian researchers.
Okuma said his project team would like to get access to more camera angles at NHL rinks than are currently available on hockey broadcasts. One idea is to retrieve footage from the many private video cameras that are located at most NHL rinks.
The insights provided by the motion database aren't restricted to hockey coaches and analysts. Sports broadcasters could use it to provide a more elaborate play commentary.
The system could also help computer game and film industries create more realistic graphics, Okuma said. Eventually, the research could be used to develop robots capable of imitating human motion.
Okuma returns in December to Japan, where he intends to pursue a PhD and go back to being a fan of his favourite sport -- soccer.
www.canada.com/vancouver/news/story.asp?id=F3A4CD63-4CD3-454B-885A-EA607CE026F5