A multiplatform solution for physical exercise based on pose detection

Zbog sve manjeg prakticiranja tjelesne aktivnosti, u okviru Erasmus+ projekta INNOSID rodila se ideja o razvoju programskog rješenja koja će služiti kao potpora za gibanje i rekreaciju kod ljudi, a pogotovo kod osoba s invaliditetom (OSI). Promatrajući postojeća rješenja za tjelovježbu, koja koriste računalni vid, nisu uočena rješenja koja se prilagođavaju OSI. U sklopu ovog rada, razvijen je prototip višeplatformskog programskog rješenja prilagođenog osobama s invaliditetom. Specifično, programsko rješenje radnog naziva 'Gibalica' razvijeno je u razvojnom okruženju Flutter za iOS i Android uređaje. Gibalica koristi tehnologiju računalnog vida za raspoznavanje čovjeka i jedanaest baznih poza tjelovježbe i njihovih kombinacija u stvarnom vremenu na temelju ulaznog videa. Korisnik je u mogućnosti koristiti aplikaciju kroz tri igre: 'Trening', 'Dan i noć' i 'Ponavljanje'. Provedene su dvije faze testiranja programskog rješenja, alfa i beta testiranje. Alfa testiranje provedeno je interno u iteracijama kroz čitav razvoj, a beta testiranje u suradnji s Hrvatskom zajednicom za Down sindrom (HZDS). Programsko rješenje dobilo je pozitivnu povratnu informaciju HZDS-a gdje se u evaluaciji programsko rješenje pokazalo motivirajuće, zanimljivo i korisno.Due to the declining practice of physical activity, within the Erasmus + project INNOSID, the idea was born to develop a software solution that will serve as a support for movement and recreation in people, especially people with disabilities (PWD). Looking at the existing exercise solutions, which use computer vision, no solutions that adapt to PWD have been observed. As part of this thesis, a prototype of a multi-platform software solution adapted for people with disabilities was developed. Specifically, the software solution working title ‘Gibalica’ was developed in the Flutter framework for iOS and Android devices. The 'Gibalica' uses computer vision technology to recognize a human and eleven basic exercise poses and their combinations in real time based on input video. The user is able to use the app through three games: ‘Training’, ‘Day and Night’ and ‘Repetition’. Two phases of software solution testing were performed, alpha and beta testing. Alpha testing was conducted internally in iterations throughout the development, and beta testing in collaboration with the Croatian Down Syndrome Association. The software solution received positive feedback from the Croatian Down Syndrome Association, where the evaluation of the software solution proved to be motivating, interesting and useful