Evaluacija metoda za segmentaciju slike zasnovanih na superpikselima

master's thesis

• Title (english): Evaluation of image segmentation methods based on superpixels
• Author: Branimir Ivić
• Mentor: Sonja Grgić (mentor)
• Committee member: Sonja Grgić (predsjednik povjerenstva)
• Committee member: Marko Bosiljevac (član povjerenstva)
• Committee member: Dario Bojanjac (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2023-07-04, Croatia

Abstract

Superpiksel je skup grupiranih susjednih piksela koji imaju slične značajke boje, luminancije i teksture. Točnije, superpikseli su kompaktne i neprekinute regije na slici s vlastitim niskorazinskim semantičkim značenjem. Njihovom primjenom smanjujemo broj ulazna i povećavamo učinkovitost postupka segmentacije. U radu se objašnjavaju osnove segmentacije slike, definicija i stvaranje superpiksela te primjena superpiksela u metodama semantičke segmentacije. Obrađuju se najčešće korišteni algoritmi superpiksela i izvodi se evaluacija istih nad poznatim skupom slika za testiranje pod nazivom NYUv2. Rad završava sa zaključkom o široj primjeni algoritama superpiksela u grani računalnog vida.

Abstract (english)

A superpixel is a set of grouped neighboring pixels that have similar color, luminance, and texture features. Specifically, superpixels are compact and continuous regions in an image with their own low-level semantic meaning. By applying superpixels, we reduce the number of inputs and increase the efficiency of the segmentation process. The paper explains the basics of image segmentation, the definition and creation of superpixels, and the application of superpixels in semantic segmentation methods. The most commonly used superpixel algorithms are covered and their evaluation is performed on a well-known set of test images called NYUv2. The paper ends with a conclusion on the wider application of superpixel algorithms in the field of computer vision.

• Keywords: računalni vid; segmentacija slike; superpikseli; konvolucijske neuronske mreže; SEEDS; SLIC; ERS; ETPS; CRS; FCN; NYUv2
• Keywords (english): computer vision; image segmentation; superpixels; convolutional neural networks; SEEDS; SLIC; ERS; ETPS; CRS; FCN; NYUv2
• Language: croatian
• URN:NBN: urn:nbn:hr:168:123839
• 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: 2023-10-11 09:06:22