Semantička segmentacija prirodnih scena dubokim neuronskim mrežama

master's thesis

• Title (english): Semantic segmentation of natural scenes with deep neural networks
• Author: Ivan Borko
• Mentor: Siniša Šegvić (mentor)
• Committee member: Siniša Šegvić (predsjednik povjerenstva)
• Committee member: Zoran Kalafatić (član povjerenstva)
• Committee member: Jan Šnajder (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2015-07-16, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

U ovom radu opisane su duboke neuronske mreže, s posebnim naglaskom na konvolucijske neuronske mreže. Detaljno su obrađene metode i postupci koji se koriste kod učenja konvolucijskih neuronskih mreža. Implementirana je konvolucijska mreža za semantičku segmentaciju u Theano radnoj okolini i isprobana na dva skupa podataka: Stanford Background skupu podatka koji za ulaz ima RGB slike i na KITTI skupu podataka koji na ulazu ima RGB-D slike. Na kraju su prikazani dobiveni rezultati, s opisima i slikama.

Abstract (english)

This work describes deep neural networks, with emphasis on convolutional neural
networks. Gradient methods and learning algorithms used for neural networks lear-
ning are shown. Convolutional network for semantic segmentation problems is built
using Theano framework and it’s performance is evaluated on two datasets: Stanford
Background dataset, consisting of only RGB images and KITTI dataset, consisting of
RGB-D images. Following that, numerical results are presented, together with des-
criptions and images.

• Keywords: računalni vid; neuronske mreže; konvolucijske mreže; duboko učenje; semantička segmentacija
• Keywords (english): computer vision; neural networks; convolutional networks; deep learning; semantic segmentation
• Language: croatian
• URN:NBN: urn:nbn:hr:168:099818
• 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: 2018-02-06 07:57:48