Inkrementalno učenje modela za raspoznavanje slike uz dodavanje novih razreda

undergraduate thesis

• Title (english): Class-incremental learning for image recognition
• Author: Dario Deković
• Mentor: Siniša Šegvić (mentor)
• Committee member: Siniša Šegvić (predsjednik povjerenstva)
• Committee member: Petra Bevandić (član povjerenstva)
• Committee member: Matej Grcić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2021-07-07, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Inkrementalno učenje uz dodavanje novih razreda jedna je od grana dubokog učenja. Nadalje, posebno je zanimljiva primjena inkrementalnog učenja uz dodavanje novih razreda na problemu raspoznavanja slika. Inkrementalno učenje uz dodavanje novih razreda izrazito je korisno u sustavima za rad u stvarnom vremenu kao što su, na primjer, roboti. Jedna od novijih metoda u tom području je konsolidacija dubokih modela i ona je detaljnije proučena i implementirana u ovom radu. Implementacija metode testirana je na dobro poznatom skupu podataka CIFAR-100 za različite konfiguracije problema.

Abstract (english)

Class-incremental learning is one of the branches of deep learning. Furthermore, the application of class-incremental learning to the problem of image recognition is particularly interesting. Class-incremental learning is very useful in real-time systems such as, for example, robots. One of the newer methods in this area is the deep model consolidation, and it has been studied in more detail and implemented in this paper. The implementation was tested on the well-known data set CIFAR-100 for different problem configurations.

• Keywords: inkrementalno učenje uz dodavanje novih razreda; računalni vid; duboko učenje; konsolidacija dubokih modela; raspoznavanje slika
• Keywords (english): class-incremental learning; computer vision; deep learning; deep model consolidation; image recognition
• Language: croatian
• URN:NBN: urn:nbn:hr:168:361687
• Study programme: Title: Computing; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva (sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-01-14 20:16:55