Razvoj sistema za prepoznavanje obrazov

magistrsko delo

Nejc Čelik (Author), Mirjana Kljajić Borštnar (Mentor)

Abstract

V magistrskem delu smo opisali problem razvoja sistemov za prepoznavanje obrazov. Sistemi za prepoznavanje obrazov morajo poleg ločevanja med obrazi različnih oseb za mnoge primere uporabe, kot je na primer kontrola dostopa, biti sposobni ločiti med resničnimi in lažnimi obrazi (npr. natisnjena fotografija). Brez sposobnosti zaznave lažnih obrazov imajo sistemi za prepoznavanje obrazov veliko ranljivost z vidika varnosti. Predstavili smo moderne tehnologije strojnega učenja, ki omogočajo izdelavo modelov za 2D prepoznavanje obrazov in prepoznavanje lažnih obrazov. Poleg tehnologij smo predstavili tudi proces razvoja modela za prepoznavanje obrazov z omejeno količino podatkov brez uporabe etično vprašljivih podatkovnih množic in proces razvoja modela za prepoznavanje lažnih obrazov.

Keywords

prepoznavanje obrazov;konvolucijske nevronske mreže;razvoj sistemov;

Data

Language:	Slovenian
Year of publishing:	2023
Typology:	2.09 - Master's Thesis
Organization:	UM FOV - Faculty of Organizational Sciences
Publisher:	[N. čelik]
UDC:	004.8
COBISS:	173244163
Views:	45
Downloads:	2
Average score:	0 (0 votes)
Metadata:

Other data

Secondary language:	English
Secondary title:	Development of a facial recognition system
Secondary abstract:	This thesis discusses the development and functionality of facial recognition systems. Besides face recognition, facial recognition systems must be able to separate between live or spoof faces for many different use cases such as access control. Without a solution to prevent face anti-spoofing facial recognition systems have big vulnerabilities from a security perspective. We explore the latest advancements in deep learning technology that allow for the creating of 2D facial recognition models and face anti-spoofing models. We present a method for training facial recognition models with limited data, eliminating the need for large, ethically questionable datasets. In addition, we also present the development of a face anti-spoofing model.
Secondary keywords:	Umetna inteligenca;Univerzitetna in visokošolska dela;
Type (COBISS):	Master's thesis/paper
Thesis comment:	Univ. v Mariboru, Fak. za organizacijske vede
Pages:	VI, 69 f.
ID:	19894530