Prepoznavanje i analiza plesnih pokreta modelima dubokog učenja

master's thesis

• Title (english): Recognition and analysis of dance movements using deep learning models
• Author: Dora Doljanin
• Mentor: Marko Horvat (mentor)
• Committee member: Marko Horvat (predsjednik povjerenstva)
• Committee member: Mirjana Domazet-Lošo (član povjerenstva)
• Committee member: Tomislav Jagušt (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2023-07-07, Croatia

Abstract

U ovome radu razvijena je računalna aplikacija MoveLab za prepoznavanje plesnih pokreta s dodatnom primjenom u vježbama posture. U tu svrhu, izrađen je vlastiti skup podataka te je primijenjena LSTM mreža za izgradnju modela dubokog učenja koja na temelju niza ključnih točaka tijela može prepoznati ljudske pokrete.
U implementaciji aplikacije razvijene u sklopu ovoga diplomskoga rada korišten je okvir MediaPipe i njegova rješenja za detekciju ključnih točaka tijela i šaka. Aplikacija omogućuje unos korisničkog identifikatora, odabir aktivnosti, igranje plesne igre, izvođenje vježbi držanja, klasifikaciju pokreta, analizu rezultata, njihovu pohranu u bazu podataka te dohvaćanje vlastite statistike.

Abstract (english)

This work presents an innovative computer application MoveLab designed to accurately recognize dance movements, while also offering practical applications in posture exercises. To achieve this objective, a new dataset was created, and an LSTM network was employed to construct a deep learning model capable of accurately identifying human movements based on a sequence of body keypoints.
The implementation of this application leveraged the solutions provided by the MediaPipe framework for detecting body and hand keypoints. The developed application provides users with a wide range of functionalities, including the ability to input a unique user identifier, select various activities, engage in interactive dance games, perform posture exercises, classify different movements, obtain the results of the executed activity, store the outcomes in a database, and access individualized statistics.

• Keywords: klasifikacija pokreta; plesni pokreti; vježbe za posturu; okvir MediaPipe; programski jezik Python
• Keywords (english): movement classification; dance movements; posture exercises; MediaPipe framework; Python programming language
• Language: croatian
• URN:NBN: urn:nbn:hr:168:561898
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-10-11 08:43:42