Johns Hopkins University engineering students Xiaojian Sun and Kevin Wu had no idea how many others had failed at what they were about to attempt — or even that any others had tried at all.
When Sun (who goes by Jason) and Wu took on a project to streamline the process for measuring bats, all they knew was that their backgrounds in computer vision were key to making it happen. That it might help the Baltimore Orioles added excitement and incentive.
The two are students at Hopkins’ Whiting School of Engineering; Sun is a first-year graduate student, and Wu is a junior. Last fall, they joined the Sports Analytics Research Group (SARG), run by Dr. Anton (Tony) Dahbura.
Over the course of the 2024-25 school year, Wu and Sun figured out how to use AI and computer vision to measure a bat’s dimensions in a matter of seconds. This could save the Orioles analytics team countless hours and manpower, as measuring a bat manually with calipers takes anywhere from an hour to an afternoon.
The Wish List
Sig Mejdal has a wish list. It probably looks different than the average fan’s wish list or even manager Brandon Hyde’s. As the Orioles’ Vice President and Assistant General Manager of Baseball Operations for the team, Mejdal is the numbers guy. He was hired at the same time as Hyde and general manager Mike Elias when the team was restructuring in 2018-2019. When Mejdal joined the Orioles, he connected with Dahbura, and the two formed a partnership to benefit both the Orioles and Hopkins students.
“So, over the years, I would have this sheet, which has suggested projects. And these projects were on that sheet because I thought they would be interesting and useful for the Orioles,” Mejdal told Baltimore Fishbowl, referring to his wish list.
For pressing, high-priority needs, the Orioles use their own analysts. Items on his project wish list, however, might not be as pressing, but have potential for big impact if solved.
Mejdal also keeps the students in mind. Would the project be useful to them? Would they learn skills they’d use either in life, or in a future baseball career?
A Long-standing Relationship
Dahbura confirms that Mejdal, the Orioles, and Hopkins have a well-established and positive partnership that benefits all involved.
“We have a long-standing relationship with the Orioles, and one of the things that Sig does is he kind of keeps a running list, kind of a wish list of projects that are suitable for our students to work on,” Dahbura told Fishbowl. “These are original research projects, not exercises.”
Dahbura said a better way to measure bats had been on the list for quite some time.
“Other groups have tried and just haven’t even come close to the results that Kevin and Jason have in terms of accuracy,” he said. “They’ve been able to…hopefully convert a very tedious task into something that can be automated and performed very quickly.”
With the increasing prominence of torpedo bats — which are customized to concentrate more of the barrel’s mass around the individual player’s “sweet spot,” creating the shape resembling, you guessed it, a torpedo — bat dimensions have become front of mind, but customization of bats to players is hardly new.
Dahbura noted that players have different bats for different hitting situations, sometimes even for different pitchers they face, and torpedo bats are another form of bat variation. Orioles catcher Adley Rutschman has been using a torpedo bat all year, though the design he uses is more subtle than the ones the Yankees have been using, Mejdal said.
Regardless of bat shape or trend, Major League Baseball teams go through an enormous number of bats, and both Dahbura and Mejdal say players go through a lot of experimentation with bats of different shapes and dimensions. Measuring them manually is time-consuming and tedious. The process involves using digital calipers to measure the circumference of a bat at one-inch intervals. So, for a 34-inch bat, there are 33 marker points an inch apart. At each marker point, three measurements are taken around the bat with the calipers and recorded.
Young Men with (Computer) Vision
When Sun and Wu joined Dahbura’s Sports Analytics Research Group at Hopkins, the bat measuring project stood out to them because they both have backgrounds in computer vision.
“Unlike other projects, this one is more about computer vision, instead of sports analysis,” Sun said.
They didn’t know until after choosing the project that other student teams had tried and failed to solve the problem.
“They [previous students] kind of achieved around 70% precision, which is not enough for what the Orioles want,” Sun said. “They want a high precision, like 99.7%, and many students saw that coming and they’re kind of scared.”
Sun and Wu found the topic interesting, though, and decided to tackle it.
They admit to learning from the less successful attempts that came before theirs, but only in the most general sense. Dahbura did not give Sun and Wu access to data or notes from previous project attempts or steer them towards one approach or another. He simply told them that previous teams had used deep learning methods and were unable to achieve accuracy higher than 70%, which was not enough for the Orioles’ purpose.
That confirmed their initial interest in using their computer vision backgrounds as the starting point for solving the problem, rather than a deep learning approach. Their resulting program operates with 99.8% efficiency, and the process is as simple as hanging a bat with fishing wire in front of a green screen, taking a photo with an iPhone and having results in under a minute.
Scouting Report
Mejdal and the Orioles analysts liked what they saw. He gives all the credit to Dahbura and the students for the energy and intellect they have and continue to put into the project.
“What was impressive is the first time they came with the question for us, they were almost done with the entire thing,” Mejdal said. “So, from just the initial specs, their internal motivation, their intelligence, their pragmatism, they solved the problem which I know others had failed at, and they did it with no input from us, other than the original specs.”
They plan to bring Sun and Wu to Camden Yards to replicate the bat-measuring set-up in a more durable way, perhaps using a harness to secure the bat instead of fishing wire.
Dahbura sees many applications for the students’ work, including in the minor leagues and beyond.
“I’m very grateful to Sig and the Orioles for giving the students this opportunity to work on something that would turn out to be so significant,” he said. “Measuring baseball bats is going to be as much a part of the daily drill of baseball teams that rubbing the baseballs with mud and chopping up the batter’s box [are]. ‘Got to go measure the bat as well.’”
On Deck
Mejdal gets to cross an item off that wish list, with the knowledge he’s provided opportunities for college students to advance their educations and the Orioles to advance their methods. Sun and Wu are preparing a paper on their work to submit to the 2026 MIT Sloan Sports Conference as their bat-measuring program advances to the next stage with the Orioles.
Once finals are over, Sun and Wu are looking forward to enjoying some outside time.
“Really my passion, it’s always been skiing, and I’m also very passionate about surfing, so that’s something else that I can look forward to for the summer,” Wu said.
“Sometimes I go kayaking, fishing around Loch Raven,” Sun said. “It’s a great place to fish and to go kayaking.”
In addition to majoring in computer science and economics as an undergraduate at Berkeley, Sun also studied music and was an opera singer.
“When I got here, actually, at first I was thinking about Peabody. But then I was like, ‘Yeah, sure, I probably can combine some of what I did, and then follow my music dream,’” Sun said.
As the old opera saying goes, “It ain’t over ’til the bat lady swings.”
