Sustav za pohranu i pregled vremenskih serija kostura ljudskog tijela

undergraduate thesis

• Title (english): System for storage and replay of human skeleton time series
• Author: Matija Rebuša
• Mentor: Siniša Popović (mentor)
• Committee member: Siniša Popović (predsjednik povjerenstva)
• Committee member: Marko Šarlija (član povjerenstva)
• Committee member: Ivan Kesedžić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2021-02-03, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Cilj projekta je opisati procese i postupke računalnog vida potrebne za snimanje i pregled vremenskih serija kostura ljudskog tijela koristeći dubinsku kameru te njihovu primjenu za prepoznavanje ljudskih gesta u strojnom učenju. Također je potrebno opisati i predložiti računalne alate korištene za implementaciju primjera takvog sustava Snimanje gesta se postiže analizom snimaka IR i RGB senzora dubinske kamere te njihovom obradom za procjenu pozicija kostura. Prostorno-vremenske koordinate glavnih dijelova kostura se zapisuju unutar tekstualnih datoteka pripadnih gesta. Pregled vremenskih serija se zatim postiže reprodukcijom pokreta kostura iz pripadnih tekstualnih datoteka gesta. Implementirana je Java aplikacija za snimanje i pohranu podataka iz vremenske serije snimaka pozicija kostura u obliku tekstualnih dokumenata te reprodukciju tih snimaka na skočnom Java Swing prozoru koristeći OpenGL. Predložene se ideje za daljnji razvoj sustava strojnim učenjem u drugim granama umjetne inteligencije.

Abstract (english)

The aim of the project is to describe the processes and procedures of computer vision required for recording and reviewing a time series of human skeleton snapshots using a depth camera and their application for human gesture recognition in machine learning. It is also necessary to describe and suggest computer tools used to implement examples of such a system. Gesture recording is achieved by analyzing images of IR and RGB depth camera sensors and processing them to assess the position of the skeleton. The spatio-temporal coordinates of the main parts of the skeleton are recorded within the text files of the corresponding gestures. An overview of the time series is then achieved by reconstructing the skeleton from the corresponding gesture text files. A Java application has been implemented to record and store data from a time series of skeleton position snapshots in the form of text documents and to reproduce these snapshots in a pop-up Java Swing window using OpenGL. Ideas for further development of machine learning systems in other branches of artificial intelligence are proposed.

• Keywords: geste; snimanje; reprodukcija; implementacija; dubinska kamera
• Keywords (english): gestures; recording; reproduction; implementation; depth camera
• Language: croatian
• URN:NBN: urn:nbn:hr:168:668494
• Study programme: Title: Computing; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva (sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-01-13 13:41:53