Stvaranje i detekcija videozapisa primjenom tehnologije DeepFake

master's thesis

• Title (english): Creation and Detection of Video Sequences Using DeepFake Technology
• Author: Antonio Kuzminski
• Mentor: Tomislav Hrkać (mentor)
• Committee member: Tomislav Hrkać (predsjednik povjerenstva)
• Committee member: Dejan Škvorc (član povjerenstva)
• Committee member: Zoran Kalafatić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2022-07-04, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Deepfake označava umjetno stvoren videozapis ili sliku koja sadrži
zamijenjeno lice a stvorena je nekim algoritmom strojnog učenja.
U ovome radu obrađeno je stvaranje deepfake sadržaja koristeći
arhitekturu duboke neuronske mreže naziva autoenkoder. Prikazani
su koraci konstrukcije skupa podataka, učenja i eksploatacije
modela. Navedene su korisne primjene deepfakea, ali isto tako i
loše primjene koje su bila motivacija za stvaranjem modela
koji detektiraju deepfake sadržaj. Drugi dio rada obrađuje zanimljiv
pristup detekciji deepfakea koristeći MRI slike lica. Na kraju
je dat zaključak i navedena područja na kojima je moguće poraditi
i time poboljšati rezultate rada.

Abstract (english)

Deepfake is an artificially made video or a picture which contains
swapped face and is made by some machine learning algorithm.
First part of the thesis is about the deepfake content creation
using the deep neural networks called autoncoder. Steps on how to
construct deepfake dataset, how to train the model and how
to make inference on the model are shown. Useful uses of the
deepfake technology are presented but also bad uses which
served as an inspiration to develop deepfake detection models.
Second part of the thesis uses interesting approach using MRI images
to detect deepfake content. At the end conclusion is given with
suggestions on how to improve results of the thesis in the future.

• Keywords: deepfake; autoenkoder; duboke neuronske mreže; MRI; GAN
• Keywords (english): deepfake; autoencoder; deep neural networks; MRI; GAN
• Language: croatian
• URN:NBN: urn:nbn:hr:168:489247
• 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: 2023-01-13 15:23:17