【香港人工智能初创公司Miro使用计算机视觉跟踪比赛中的运动鞋】

【香港人工智能初创公司Miro使用计算机视觉跟踪比赛中的运动鞋】Miro首席执行官Taylor Host曾在美国领导麦当劳项目,利用计算机视觉帮助快餐连锁店通过提高厨房效率和测量配送时间来销售更多汉堡。利用同样的技术,Miro公司推出了根据运动服和鞋子识别和收集马拉松跑步者数据的视觉跟踪产品,并首次在苏州马拉松大赛中亮相,不仅帮助运动员获得自己比赛的图像和数据,还为赛事赞助商提供市场数据以改进营销策略。

The Miro system can scan the thousands of running shoes in a marathon and supply the data to shoe manufacturers. Photo: Handout

Miro chief executive Taylor Host once led projects for McDonald’s in the US, using computer vision to help the fast food chain sell more burgers by improving kitchen efficiency and measuring drive-through times.

Today, Host’s artificial intelligence (AI) start-up Miro uses the same technology but in a very different way – to identify and collect data on runners in marathons, including the sportswear and shoes they wear.

“We’re trying to make computers use cameras the same way that a human uses their eyes,” said Host. “Whether you want to identify an automobile or a face, computer vision is one of those areas of artificial intelligence where we’re seeing a lot of rapid advancement.”

Hong Kong-based Miro is one of many AI start-ups in the region which are developing computer vision technology. SenseTime, a Hong Kong and Beijing-headquartered company and the world’s most valuable AI start-up at US$4.5 billion, is one of the global leaders in computer vision. Chinese AI company Megvii is also another firm in China that is known for its facial recognition technology Face++.

But while companies like SenseTime and Megvii have largely focused on facial recognition, Miro has trained its computer vision model to identify numbers and objects, such as bib numbers – a “licence plate” of sorts – that each marathon runner wears during a race, and the brand logos of their shoes and running gear.

For example, Miro can use its computer vision model to process thousands of photographs from a race and pick out individual racers from their bib number. This allows racers to easily find pictures of themselves via their individual bib number.

Host and his team at Miro may have struck gold by choosing to concentrate on marathons and athletes. In China, running events have become a booming industry as the country’s white-collar workers see long-distance running as a fitness status symbol.

The number of marathons organised in China has nearly quadrupled, from over 300 races in 2016 to 1,102 in 2017, according to Statista data. In 2017, almost 5 million participants took part in marathons.

“China is a huge market, and it’s desirable for anyone in the industry to be working in China because of that growth,” said Host, who pointed out that Southeast Asia was also a fast-growing running market.

Miro works with a variety of marathon organisers across the US and Asia, including China, where it recently executed its technology at a recent marathon this month in Suzhou. The company works directly with race organisers, for example, to obtain access to high-resolution photographs captured by the hundreds of professional photographers who take pictures of participants as they run the course.

Aside from helping race participants find pictures of themselves, major shoe and sportswear brands, as well as race sponsors are often interested in sussing out market insights. Data such as how many participants are wearing a particular brand or model of shoe can help companies like Nike and Adidas refine marketing strategies.

Taylor Host, chief executive of Hong Kong-based artificial intelligence start-up Miro, hopes to use computer vision technology to help the sportswear brand gain better customer insights. Photo: SCMP

Brands and retailers are always interested to “understand where they sit in a particular market … and the uptake of [brands] over an event or a set of events”, Host said. Miro’s models are now trained to identify some 19 brands of sports shoes, from Nike to China brands such as Anta and Li-Ning.

Currently, its computer vision technology can even assess the amount of wear on a pair of shoes, which is valuable information as brands often want to know when people are ready to replace their current pair of running shoes.

Host was quick to point out that while Miro has access to data provided by race organisers, it is anonymised and that organisers are typically the ones which disclose to participants how photos and videos taken of them are used.

“Everything that we do is tied to that number and not to an individual identity,” he said. “We do not know who you are, we merely assign data that we extract to that bib number.”

Moving forwards, Miro hopes to do more around gait recognition with computer vision, for example by allowing the technology to analyse the way athletes run to improve their performance.

The company has partnered with medical professionals and organisations both in China and globally, to explore how computer vision can help with physiotherapy and gait analysis to improve an athlete’s training while assisting them in avoiding injury.

Host, who played baseball in the US as a teenager and all through college, has had experience with a computer vision system which helped him identify that he needed to stop throwing a particular type of pitch to avoid permanent injury that would prematurely end his baseball career.

“Computer vision was a part of my sports career and now it’s about finding an opportunity to do the same thing, for another sport,” he said.

https://www.scmp.com/tech/enterprises/article/3003522/big-macs-nikes-hong-kong-ai-start-miro-uses-computer-vision-track


Comments are closed.



无觅相关文章插件