Integracija modela difuzije u vizualno odgovaranje na pitanja

master's thesis

• Title (english): Integrating diffusion models into visual question answering
• Author: Dora Ivezić
• Mentor: Marina Bagić Babac (mentor)
• Committee member: Marina Bagić Babac (predsjednik povjerenstva)
• Committee member: Vedran Mornar (član povjerenstva)
• Committee member: Krešimir Fertalj (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2024-07-11, Croatia

Abstract

Vizualno odgovaranje na pitanja (engl. Visual Question Answering - VQA) je automatizirano
rješenje za odgovaranje na pitanja u vezi sa sadržajem slike. Cilj ovog rada je osmisliti, implementirati i evaluirati rješenje ovog izazova korištenjem difuzijskog modeliranja, uvjetujući ga značajkama slike i prikazima pitanja. Istražuje se u kojoj mjeri integracija difuzijskih modela može poboljšati kvalitetu generiranih odgovora. Također, napravljena je usporednu analizu između predloženog rješenja i drugih najsuvremenijih VQA metoda.

Abstract (english)

Visual Question Answering (VQA) is an automated solution for responding to questions regarding the content of an image. The objective of this thesis is to design, implement, and evaluate a solution to this challenge using diffusion modeling, conditioning it with image features and question representations. This research investigates to what extent the integration of diffusion models can improve the quality of generated responses. In addition, it includes a comparative analysis between the proposed solution and other state-of-the-art VQA methods.

• Keywords: difuzijsko modeliranje; vizualno odgovaranje na pitanja (VQA)
• Keywords (english): Diffusion Model; Visual Question Answering (VQA)
• Language: english
• URN:NBN: urn:nbn:hr:168:139070
• 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: 2024-10-12 20:52:55