Raspoznavanje slika učenjem sličnosti

master's thesis

• Title (english): Image classification by learning similarity
• Author: Dorian Farkaš
• Mentor: Siniša Šegvić (mentor)
• Committee member: Siniša Šegvić (predsjednik povjerenstva)
• Committee member: Jan Šnajder (član povjerenstva)
• Committee member: Adrian Satja Kurdija (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2024-07-10, Croatia

Abstract

Raspoznavanje slika je ključni problem računalnog vida s mnogim primjenama. Najbolji rezultati postižu se dubokim modelima s konvolucijama i slojevima pažnje, posebno kada je broj razreda velik ili nepoznat. U ovom radu odabrani su alati za automatsku diferencijaciju i analizirane duboke arhitekture za učenje sličnosti. Slobodno dostupni i novoizgrađeni skupovi slika podijeljeni su na podskupove za učenje, validaciju i testiranje te su prikazani rezultati eksperimenata, uz preporuke za budući rad.

Abstract (english)

Image recognition is a key problem in computer vision with many applications. The best results are achieved with deep models using convolutions and attention layers, especially when the number of classes is large or unknown. This work selected tools for automatic differentiation and analyzed deep architectures for similarity learning. Freely available and newly constructed image datasets were divided into subsets for training, validation, and testing, and experimental results were presented, with recommendations for future work.

• Keywords: raspoznavanje slika; računalni vid; duboko učenje; konvolucijske neuronske mreže
• Keywords (english): image recognition; computer vision; deep learning; convolutional neural networks
• Language: croatian
• URN:NBN: urn:nbn:hr:168:707694
• 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: 2024-10-12 20:34:01