Seizing an opportunity from the new wave of in-home workouts, Hong Kong-based fitness solutions technology startup OliveX has developed a smart fitness mirror that deploys artificial intelligence (AI) to detect users’ body movements, and provide real-time feedback and performance analysis.
Mounted on a wall or placed on a stand, “Kara” is an actual mirror that can be used for the morning routine, but with technology built inside.
Equipped with speakers, camera, and software powered by a processor, it tracks and analyzes body movements with AI technology, and detects nuances in limb movements, allowing people to correct their form during the exercise routine.
The system provides instant feedback and instructions, just like a personal training coach.
OliveX also launched a fitness app Lympo Squat, which, too, is powered by AI technology and enables users to perform squats and track their performance through a smartphone camera.
There are built-in fitness menus for Kara, such as yoga, pilates, high-intensity interval training (HIIT), among others. It also allows users to connect with monitoring devices and join livestream classes.
Still in the test phase, OliveX has partnered with instructors to provide live-streaming fitness classes from 7 pm to 8 pm every day.
Prior to the fitness technology venture, OliveX’s co-founder Keith Rumjahn had been involved in Coachbase, a well-known basketball coaching app.
Kara is designed to allow people to work out from the comfort of their home, Rumjahn told the Hong Kong Economic Journal.
A “home” edition of Kara to cater to individual users working out at home comes at a lump-sum price of HK$10,000, or a monthly fee of HK$800 for a rental option.
Meanwhile, a “commercial” edition is priced at HK$15,000, plus an additional monthly fee of HK$700 for exclusive courses.
Rumjahn revealed that his firm has sold about 20 Kara smart fitness mirrors, with major customers being hotels, clubhouses of private housing estates, and co-working spaces across the city.
In an interview, the entrepreneur said one of the major challenges in developing Kara was to access sufficient data, in order to train its pose-detection AI software.
“Relevant data to train our machine learning algorithms was just not sufficient,” he said. “Each sport involves a massive number of poses and movements, and there are also many complicated movements, say, handstand, that fewer people are able to do it and provide us the data.”
He had once called on sports enthusiasts, via social media platforms, to offer their personal training videos to the firm, but failed to get the desired response.
Now the firm has rented a studio and hired sports enthusiasts to collect recordings of their body movements in multiple sports.
Rumjahn stressed that the built-in camera of Kara does not have a video-recording feature and that users’ movements and poses will not be recorded and sent back to the firm.
To protect users’ privacy, the data is automatically deleted once Kara processes the information in real-time, he pointed out.
The full article appeared in the Hong Kong Economic Journal on Sept 18
Translation by Ben Ng
[Chinese version 中文版]
– Contact us at [email protected]