If you’ve ever been told you have a (ahem) unique style of dancing, it turns out that’s not so far off.
Researchers at the University of Jyväskylä in Finland have trained a computer to recognise a person, simply by analysing their funky moves.
The computer algorithm used motion capture to look at 21 different points on 73 participants dancing freely to eight different music genres: blues, country, dance/electronica, jazz, metal, pop, reggae and rap.
The original aim of the study was to find out if the computer could pick the style of music.
But by working out their movement patterns, the AI was able to accurately identify the correct person 94% of the time – suggesting we all have a unique ‘signature movement’ (mine is still the good ol’ ‘Gangnam Style’).
“Dance style seems to be a kind of fingerprint,” PhD student Pasi Saari, author of the machine learning section of the study, said. “Each of us has distinctive movement patterns that remain pretty much the same no matter what kind of music plays.”
The algorithm had a much harder time trying to identify the style of music – its best results were picking styles with more ‘unique’ moves like metal (53% of the time) and jazz (35% of the time).
We’re guessing head-banging and step-ball changes are a lot easier to spot.
