Artificial intelligence taught to recognize a person by dance

Finnish scientists have created a neural network to recognize a person by dance. Tests have shown that artificial intelligence is able to determine a person by movements in free dance with an accuracy of 90%.

The movements of a person during a dance often correspond to the genre of music and its tempo. This process is two-way - that is, not only music determines the dance, but also the dance fits the music.

Based on this hypothesis, the researchers asked 73 volunteers to dance to musical works of 16 different genres - from the song “Want You More!” By Duran Duran to the jazz “Muskrat Ramble”.

According to  hightech.fm, during the dance, volunteer movements were read by motion capture sensors, on the basis of which the researchers made a two-dimensional model of the body of the participants in the experiment.

Then the scientists developed an algorithm based on the support vector method and gave it the task of comparing the dance record with the one who dances it.

Recognition accuracy was extremely high - about 90%. The authors of the study note that the neural network best determined the dancer by the movements of the head, neck and shoulders.

© All rights reserved. Citing to www.ict.az is necessary upon using news