Konvolucijske reprezentacije za dohvat slike na temelju sadržaja

master's thesis

• Title (english): Convolutional representations for content-based image retrieval
• Author: Filip Jakov Bulić
• Mentor: Siniša Šegvić (mentor)
• Committee member: Siniša Šegvić (predsjednik povjerenstva)
• Committee member: Zoran Kalafatić (član povjerenstva)
• Committee member: Slobodan Ribarić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2018-07-13, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Pretraživanje slikovnih baza prema slikovnom sadržaju je važan zadatak računalnog vida. U posljednje vrijeme, veliki uspjeh u tom području ostvaruju pristupi temeljeni na naučenim značajkama do kojih dolazimo dubokim konvolucijskim modelima. U radu su opisane duboke neuronske mreže s fokusom na konvolucijske neuronske mreže. Prikazan je i implementiran model DELF u razvojnom okviru Tensorflow. Model je treniran na skupu podataka znamenitosti i evaluiran je na podatkovnom skupu Oxford5k. Na kraju su prikazani dobiveni rezultati s opisima i slikama.

Abstract (english)

Content-based image retrieval is an important task in the field of computer vision. Lately, great success in this area is acomplished using deep convolutional features. In this thesis, deep neural networks are described with emphasis on convolutional networks. The DELF model is introduced and implemented using the Tensorflow library. The model was trained on a landmark dataset and evaluated on the Oxford5k landmarks dataset. Finally, results with descriptions and images are presented.

• Keywords: dohvat slika na temelju sadržaja; duboke konvolucijske neuronske mreže; prijenos znanja; model pažnje
• Keywords (english): content-based image retrieval; deep convolutional networks; transfer learning; attention model
• Language: croatian
• URN:NBN: urn:nbn:hr:168:976512
• 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 19:26:43