Detekcija lica u slikama utemeljena na proširenim Haarovim značajkama

master's thesis

• Title (english): Face detection in images based on extended Haar features
• Author: Andrea Kurek
• Mentor: Sonja Grgić (mentor)
• Committee member: Sonja Grgić (predsjednik povjerenstva)
• Committee member: Tomislav Kos (član povjerenstva)
• Committee member: Gordan Ježić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2019-07-04, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Electrical Engineering

Abstract

Cilj rada je detaljan teorijski i praktični osvrt na tehnologiju koja u posljednje vrijeme dobiva sve veći značaj u području umjetne inteligencije – detekciju lica. Teorijski dio rada daje uvid u sam pojam detekcije lica, njenu primjenu i problematiku koja se pri tom javlja. Osim toga, objašnjeni su pojmovi važni za razumijevanje ostatka rada, primjerice omjer detekcije, te je opisana klasifikacija osnovnih metoda i podmetoda na koje se dijeli detekcija lica. Također, dana je detaljna analiza najvažnijeg algoritma koji se koristi u primjeni detekcije lica – Viola-Jones algoritma. Na temelju tog algoritma, točnije na temelju Haarovih značajki kao jedne od temeljnih faza algoritma, implementirani su detektor lica, očiju, usta, ušiju i zjenica s mogućnošću rada na unaprijed zadanim slikama i u stvarnom vremenu. Testiranjem tih detektora izračunati su omjeri detekcije na temelju kojih su doneseni zaključci o performansama pojedinih detektora.

Abstract (english)

The object of this paper is to provide a detailed theoretical and practical overview of the technology that has gained ever greater significance in the field of artificial intelligence – face detection. The theoretical part of the paper gives an insight into the very concept of face detection, its usage and problems that appear. In addition, some concepts important for understanding the rest of the paper, such as detection rate and classification of face detection methods, are illustrated. Also, a detailed analysis of the most important algorithm used in face detection, Viola-Jones algorithm, is given. Based on this algorithm and its most important part, Haar cascades, a face, eyes, mouth, ears and pupil detectors are implemented. Detectors are able to work in offline or real time mode. By testing these detectors, detection rates were calculated based on which conclusions about the performance of individual detectors were drawn.

• Keywords: detekcija lica; omjer detekcije; Viola-Jones algoritam; Haarove značajke; detektor lica; detektor očiju; detektor usta; detektor ušiju; detektor oblika (zjenica)
• Keywords (english): face detection; detection rate; Viola-Jones algorithm; Haar cascades; face detector; eyes detector; mouth detector; ears detector; pupil detector
• Language: croatian
• URN:NBN: urn:nbn:hr:168:049948
• Study programme: Title: Information and Communication Technology; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka informacijske i komunikacijske tehnologije (magistar/magistra inženjer/inženjerka informacijske i komunikacijske tehnologije)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2020-01-07 19:33:34