Primjena neuronskih mreža u obradi slika

master's thesis

• Author: Tomislav Levanić
• Mentor: Saša Singer (mentor)
• Committee member: Saša Singer (predsjednik povjerenstva)
• Committee member: Tina Bosner (član povjerenstva)
• Committee member: Vedran Čačić (član povjerenstva)
• Committee member: Josip Tambača (član povjerenstva)
• Granter: University of Zagreb; Faculty of Science; (Department of Mathematics); Zagreb
• Defense date and country: 2018-03-02, Croatia
• Scientific / art field, discipline and subdiscipline: NATURAL SCIENCES; Mathematics

Abstract

Konvolucijske neuronske mreže (CNN) predstavljaju popularan model neuronskih mreža, izrazito pogodan za obradu slika. U današnje vrijeme, gotovo sve state-of-the-art mreže za obradu slika baziraju se na ovoj arhitekturi. Prvi dio ovoga rada definira i proučava matematički model konvolucijskih neuronskih mreža. Preciznije, definiraju se slojevi potrebni za konvolucijski model te algoritam propagiranja unazad. U drugome dijelu daje se primjena te implementacija konvolucijskog modela mreže u obliku programa za klasifikaciju slika. Implementacija je izvedena unutar programskog jezika Python te se bazira na modulu tensorflow.

Abstract (english)

Convolutional neural networks (CNN) represent a popular neural network model, extremely suitable for image processing. At present time, almost every state-of-theart neural network for image processing is based on this architecture. The first part of this thesis defines and researches a mathematical model of convolutional neural networks. More precisely, we define various layers used inside a convolutional model and the backpropagation algorithm. The second part deals with CNN implementation and application for image based classification. The programming language Python and its module tensorflow are used for the implementation.

• Keywords: konvolucijske neuronske mreže; CNN; obrada slika; algoritam propagiranja unazad; Python
• Keywords (english): convolutional neural networks; CNN; image processing; backpropagation algorithm; Python
• Language: croatian
• URN:NBN: urn:nbn:hr:217:127571
• Study programme: Title: Computer Science and Mathematics; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra računarstva i matematike (magistar/magistra računarstva i matematike)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Open access
• Created on: 2018-09-04 11:45:46