Učenje dubokih korespondencijskih metrika trojnim gubitkom

master's thesis

• Title (english): Learning deep correspondence metrics with triplet loss
• Author: Marin Kostelac
• Mentor: Siniša Šegvić (mentor)
• Committee member: Siniša Šegvić (predsjednik povjerenstva)
• Committee member: Jan Šnajder (član povjerenstva)
• Committee member: Tomislav Hrkać (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2018-07-19, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Stereoskopska rekonstrukcija je bitno polje računalnog vida. Najbitniji dio problema je pronalazak korespondentnih piksela u dvjema slikama iste scene. To je moguće ostvariti koristeći korespondencijske metrike. Pri učenju korespondencijskih metrika od velike pomoći je trojni gubitak. Definiranjem odnosa med̄u podacima moguće je naučiti metriku koja olakšava pronalazak korespondentnih piksela. U okviru rada opisan je problem stereoskopske rekonstrukcije i primjena korespondencijskih metrika i trojnog gubitka na njega. Opisan je i testiran model za rješavanje opisanog problema.

Abstract (english)

Stereoscopic reconstruction is an important field in computer vision. The most important part of the problem is finding the corresponding pixels in two images of the same scene. It is possible to achieve that using correspondence metrics. Triplet loss is a big help in training correspondence metrics. By defining the data relations it is possible to train a metric which makes finding corresponding pixels easier. in this thesis, the stereoscopic reconstruction problem is described. Application of correspondence metrics and triplet loss to it are explained. A model for solving the problem is described and tested.

• Keywords: stereoskopska rekonstrukcija; korespondencijske metrike; trojni gubitak; konvolucijske mreže
• Keywords (english): stereoscopic reconstruction; correspondence metrics; triplet loss; convolutional networks
• Language: croatian
• URN:NBN: urn:nbn:hr:168:039533
• 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: 2019-03-13 21:45:09