Peter Szirmak isn’t much of a tennis player. He enjoys the game recreationally but has certainly never approached professional status, and, at 44, isn’t likely to, either. Off the court, though, Szirmak is a nimble technology entrepreneur who made a fortune on the belief that Y2K would topple the world’s corporate infrastructure. “I tell my wife I have a good nose for what makes money,” he says, not immodestly.
Szirmak’s latest venture combines tennis and technology. A computerized line-calling system, called Auto-Ref, it uses high-speed cameras and software algorithms to determine the location of a ball within four millimetres. “Auto-Ref really elegantly replaces what we as humans see,” says Szirmak. If adopted, the product could potentially eliminate the need for linesmen and forever change the way we watch tennis on TV, with more accurate replays and a wealth of new statistical data with which to enthrall fans (and bore others).
Still in its formative days, Auto-Ref has proven it can work. In November, it received a U.S. patent for its optical line monitor for tennis. That same month, it was successfully demonstrated and tested by the International Tennis Federation, which oversees top-flight matches, at the Bell Challenge in Quebec City. During the same tournament, Auto-Ref provided sports broadcaster TSN with animated video replays for the finals. And with plenty of interest shown last month by organizers of the Australian Open, it has become a matter of when, not if, tennis will adopt some form of technology to assist officials. The call could come in the next couple of months.
Yet, despite Auto-Ref’s potential and Szirmak’s track record of success with other technologies developed while at Information Balance Inc. â a Toronto IT shop he co-founded in 1987 and still works at today â he finds himself in the same position as thousands of other Canadian entrepreneurs: he needs money to turn his dream into a winning play. Szirmak has largely been funding Waterloo, Ont.-based Auto-Ref Inc. since its inception, in 2000, with the money he made from fixing companies’ old computer systems to handle the Y2K date change, although he’s been looking for outside backing for the past three-and-a-half years. He’s tried venture capital, banks, and government aid, but, so far, hasn’t scored many points. He has managed to get some small angel investors, but nothing major.
On the surface, Szirmak shouldn’t have any trouble. He has a product that works, interest from his target market, and a variety of other sports to capitalize on once tennis is onboard. Meanwhile, the venture capital industry has $4.9 billion to spend â on top of the $1.8 billion it shelled out last year â and one insider with the Canadian Innovation Centre, a non-profit resource for budding inventors based in Waterloo, estimates there are roughly 400 public organizations that are supposed to fund entrepreneurs and/or small businesses as part of their mandate.
Everybody talks a good game, but finding money for innovation in Canada is still difficult. “The financing market kind of settled in the early part of the 21st century, and angel investors became venture capitalists, venture capitalists became bankers,” Szirmak says. “VCs will only talk to you if revenues are coming in, contracts are signed and everything, so there’s not a heck of a lot of venture about it.”
Szirmak’s failure to raise funding hasn’t been for lack of trying. He’s talked with several private and public financiers, including the Business Development Bank of Canada, but, as he points out, they get thousands of applicants every year who think they’re going to change the world. What makes Auto-Ref’s task even tougher is that potential backers tend to shy away when they find out there’s nothing to compare Auto-Ref with to judge potential revenues and pricing models. “The fact is that we probably would have had an easier time if we were a good product in an existing marketplace,” says Szirmak. “If you try to really revolutionize something, there really is no competition, and for any VC or fund, that’s a red flag.”
Professional tennis has been using a simple machine, called Cyclops, since 1980 to help officials rule on service points, but with today’s bigger and stronger male players routinely bashing the ball around at speeds of about 160 kilometres per hour (female play is about 20% slower), a much more complex electronic eye, such as Auto-Ref, is needed to cover the entire court. Convincing the conservative tennis world that the game needs a digital upgrade has proven expensive: Szirmak has burned through more than $25,000 a month on marketing and technology development. That’s pretty small potatoes compared with what his primary competitor, Hawk-Eye Innovations Ltd., has to spend. It was bought in 2002 for Â£700,000 (about $1.6 million) by Television Corp. PLC, one of the U.K.’s largest independent suppliera of TV content. Hawk-Eye â which is already used by several major TV networks, including the BBC and ESPN, to show replays during many of the world’s top tennis tourneys â is rumoured to have been financed to the tune Â£5 million (about $11.7 million) three years ago, when Auto-Ref was just getting started.
Even if the International Tennis Federation gives Auto-Ref its blessing, it could also do the same for Hawk-Eye, leading to costly marketing wars for each independently run tournament. Hawk-Eye currently has the financial clout and existing TV relationships that give it a step up â something Szirmak has tried to balance by partnering with Information & Display Systems LLC, a Florida-based company that serves up technology and administrative support at sporting events, including pro tennis. But Auto-Ref might have the better system, with its patent in the U.S. and ones for Australia and Europe pending.
Conceptually, Auto-Ref, which costs from US$25,000 to US$50,000 to install, is very simple: it takes images of a ball in flight and then determines where it lands. That’s what the human brain tries to do, but the speed at which tennis is played makes it virtually impossible for a linesman to always accurately judge. On close line calls, the human eye is right roughly 80% of the time, says Gordon McKay, Auto-Ref’s president. In other words, one out of every four close calls is going to be wrong. Using high-speed cameras to track a ball seems a simple enough solution, but cameras see things in two dimensions, and the view might be obscured by a player’s racket or body or the fact that a tennis ball actually changes shape as it lands.
Auto-Ref hired professors at the University of Waterloo to get around the obstacles. They designed a system that uses two to eight high-speed Pulnix 1024 CL cameras, connected by physical cables to their own dedicated computers to process images of the ball coming in at 120 frames per second. Before the game even begins, the cameras take controlled pictures of the court so the computers know exactly where they are relative to the court. Once the game begins, the computer looks for ball trajectories â typically when a ball is captured in six consecutive frames â and sends them to a central computer, which combines that with trajectory information from other cameras. The central computer then uses an algorithm to determine the actual trajectory of the ball in three dimensions, and sends a visual to a TV device or an umpire’s hand-held computer. The whole process takes about half a second, so umpires can quickly overrule a linesman or order a replay.
It’s the kind of thing Serena Williams would have loved to have around during the U.S. Open last September, when a series of very questionable line calls contributed to her quarter-final loss to Jennifer Capriati. “You’ll still need a couple of line judges to help you with the general stuff and to keep it honest,” says McKay. “But this is now a tool much as police have radar guns to make sure we all stay honest.”
Szirmak says there are a number of other potential markets. Auto-Ref collects all the data viewed during a game, information that can be sold and used for real-time analysis of how a player is doing or for training purposes later on. Swap in a few cheaper components, such as lesser-powered cameras, and Auto-Ref might even be cost effective for amateur tennis clubs. Szirmak has also had discussions with the International Badminton Federation, a company in New Zealand interested in using Auto-Ref in cricket, and someone in Germany who sees potential for volleyball. It’s a lot to choose from, but a nice problem to have. “You can have an outline of a strategy, but what actually happens is driven by the market,” says Szirmak.
But before any of that can happen, someone else with deep pockets will have to come out of the woodwork. Szirmak remains positive that additional financing will come through. “Given the size of our company, the challenges we’ve had and the lack of financing, we’re pretty well taking Hawk-Eye to the corner,” says Szirmak. In tennis, that’s a great place to put your opponent.
The auto-ref file
The Company Auto-Ref Inc. is a five-year-old company in Waterloo, Ont., that has developed a tennis line-calling system for determining whether balls are in or out, using between two and eight high-speed cameras and software algorithms.
The Success The Auto-Ref system is accurate to within four millimetres, which was confirmed by the International Tennis Federation last November at the Bell Challenge in Quebec City. Sports broadcaster TSN also used Auto-Ref for animated video replays for the tournament’s final match.
The Challenge Auto-Ref is in deep competition with U.K.-based Hawk-Eye, and both could be approved for use by the ITF. That could set up a marketing battle for each and every tournament. Hawk-Eye, backed by Â£5 million (about $11.7 million) in funding, would seem to have the edge over Auto-Ref in that regard. Auto-Ref needs a capital infusion, but traditional investors aren’t big on taking risks in unproven markets.