Creation and detection of malicious synthetic media - a preliminary survey on deepfakes

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216305%3A26230%2F22%3APU144472" target="_blank" >RIV/00216305:26230/22:PU144472 - isvavai.cz</a>

Výsledek na webu

<a href="https://europen.cz/Anot/54-1/sbornik-54.pdf" target="_blank" >https://europen.cz/Anot/54-1/sbornik-54.pdf</a>

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Creation and detection of malicious synthetic media - a preliminary survey on deepfakes

Popis výsledku v původním jazyce

Deepfakes present an emerging threat in cyberspace. Recent developments in machine learning make deepfakes highly believable and very difficult to differentiate between what is real and what is fake. Not only humans but also machines struggle to identify deepfakes. Current biometrics systems might be easily fooled by carefully prepared malicious synthetic media -- deepfakes. We provide an overview of deepfake creation and detection methods for selected visual and audio domains. We put specific emphasize on the open-source solutions. We discuss both facial and speech deepfakes, and for each domain, we define deepfake categories and differences between them and discuss available detection methods. For each deepfake category, we provide an overview of available open-source tools for their creation and datasets.

Název v anglickém jazyce

Creation and detection of malicious synthetic media - a preliminary survey on deepfakes

Popis výsledku anglicky

Deepfakes present an emerging threat in cyberspace. Recent developments in machine learning make deepfakes highly believable and very difficult to differentiate between what is real and what is fake. Not only humans but also machines struggle to identify deepfakes. Current biometrics systems might be easily fooled by carefully prepared malicious synthetic media -- deepfakes. We provide an overview of deepfake creation and detection methods for selected visual and audio domains. We put specific emphasize on the open-source solutions. We discuss both facial and speech deepfakes, and for each domain, we define deepfake categories and differences between them and discuss available detection methods. For each deepfake category, we provide an overview of available open-source tools for their creation and datasets.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Projekt

—

Návaznosti

S - Specificky vyzkum na vysokych skolach

Rok uplatnění

2022

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

Sborník příspevků z 54. konference EurOpen.CZ, 28.5.-1.6.2022

ISBN

978-80-86583-34-1

ISSN

—

e-ISSN

—

Počet stran výsledku

21

Strana od-do

125-145

Název nakladatele

Czech Open Systems User's Group

Místo vydání

Radešín

Místo konání akce

Radešín

Datum konání akce

29. 5. 2022

Typ akce podle státní příslušnosti

CST - Celostátní akce

Kód UT WoS článku

—