Strictly Come Dancing and Dancing with the Stars judges watch out. Artificial intelligence is taking on its next challenge: dance talent appraisal.
It sounds unlikely, but Abu Zaher Faridee and colleagues at the University of Maryland, Baltimore County, revealed their at the HotMobile conference in Arizona, last month.
Advertisement
Dancers often get in each other鈥檚 way as they make their moves and normally significantly outnumber the teacher. This makes it hard for them to assess if everyone is dancing correctly. So the team put sensor on听people learning an Indian dance called lasya. During the class, everyone wore two wrist and two ankle worn motion sensors each.
After a teaching session, the sensor data was fed into an artificial intelligence algorithm, which had been trained by an expert lasya dancer. In early tests, the system, dubbed HappyFeet, recognised when the student was dancing in the lasya style 94 per cent of the time and could give them a score as to how closely they had followed it.
The team now face a much bigger challenge: getting HappyFeet working on cheap sensors like FitBits and generalising the AI to recognize other dance styles like modern, tap, ballet, jazz, street and ballroom. Ultimately, says Faridee, they want it to be a smartphone app that acts as 鈥渁 personalised dance tutor鈥�.
With doom-laden portents about artificial intelligence coming thick and fast, Faridee sees HappyFeet as showing another side. 鈥淎I can be much more fun than it gets credit for,鈥� he says.
Read more:听Waltzing robot teaches beginners how to dance like a pro
听
