Izračun trodimenzionalne ljudske poze iz više pogleda

master's thesis

• Title (english): Multi-view 3D Pose Estimation
• Author: Luka Kušina
• Mentor: Tomislav Pribanić (mentor)
• Committee member: Tomislav Pribanić (predsjednik povjerenstva)
• Committee member: Marko Subašić (član povjerenstva)
• Committee member: Željka Lučev Vasić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2022-02-18, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

U sklopu diplomskog rada pokazali smo dosadašnje pristupa procjene ljudske poze u
dvije ili tri dimenzije i iz jednog ili više pogleda. Predstavili smo nekoliko razlicitih pristupa ˇ
procjene ljudske poze te naveli njihove prednosti i mane u odnosu jedni na druge. Posebnu
pozornost posvetili smo pristupima procjene trodimenzionalne ljudske poze iz više pogleda.
Potom smo napravili uvod u korištene tehnologije i podatkovni skup Human3.6M. Predstavljene su dvije metode procjene trodimenzionalne poze ljudskog tijela iz više pogleda te su
detaljnije razradene zajedno sa objašnjenjem poteško ¯ ca koje su se manifestirale u sklopu ´
rada.

Abstract (english)

In the scope of this master’s thesis, we have shown recent attempts of two-dimensional
and three-dimensional human pose estimation from one view or from a multi-view perspective. We introduced several different approaches to human pose estimation and we compared their strengths and weaknesses compared to one another. Special focus was given
to three-dimensional human pose estimation. Afterwards we made an introduction to the
technologies we used and to the Human3.6M dataset. Two methods of assessing the three-dimensional posture of the human body from several perspectives are presented and elaborated in more detail together with an explanation of the difficulties that manifested within the
work.

• Keywords: Procjena ljudske poze; Trijangulacija; Human3.6M; Duboko učenje; Neuronske Mreže; Računalni vid; PyTorch
• Keywords (english): Human pose estimation; Triangulation; Human3.6M; Deep Learning; Neural Networks; Computer Vision; PyTorch
• Language: croatian
• URN:NBN: urn:nbn:hr:168:000042
• 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: 2023-01-13 15:22:44