Procjena dobi analizom slika lica

undergraduate thesis

• Title (english): Age estimation using face images analysis
• Author: Nikola Perić
• Mentor: Sonja Grgić (mentor)
• Committee member: Sonja Grgić (predsjednik povjerenstva)
• Committee member: Jelena Božek (član povjerenstva)
• Committee member: Ivana Žeger (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2024-07-08, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

U ovom radu, objašnjene su neke od glavnih metoda prepoznavanja dobi iz slika lica osobe kao što su linearna diskriminantna analiza, metoda k-najbližih susjeda, konvolucijske neuronske mreže, analiza glavnih komponenti i stroj s potpornim vektorima. Implementiran je model koji koristi analizu glavnih komponenti kako bi smanjio broj dimenzija sustava i zatim koristi stroj s potpornim vektorima kako bi klasificirao kojoj klasi tj. dobnoj skupini pripada osoba. Kao baza podataka koristi se UTKFace skup podatka koja sadrži oznake dobi, spola i etničke pripadnosti osobe. Implementirani model postiže točnost od 48.88 % te je, prema matrici zabune vidljivo da algoritam ima probleme s rubovima klasa što sugerira da bi se zadatku moglo pristupiti kao problemu regresije umjesto klasifikacije.

Abstract (english)

In this paper, some of the main methods for detecting the age of a person based on images are explained, such as linear discriminant analysis, the k-nearest neighbors method, convolutional neural networks, principal component analysis and support vector machines. A model was implemented that uses principal component analysis for dimensionality reduction of the dataset, after which a support vector machine is used to classify which class (age group) a person belongs to. As a database, the UTKFace dataset is used, which contains the labels for age, gender, and ethnicitiy of the person. The implemented model achieves an accuracy of 48.88 % and, based on the confusion matrix, it is evident that the algorithm struggles with the class edges, which suggests that the task could be approached as a regression problem instead of classification.

• Keywords: linearna diskriminantna analiza; metoda k-najbližih susjeda; konvolucijske neuronske mreže; analiza glavnih komponenti; stroj s potpornim vektorima
• Keywords (english): linear discriminant analysis; k-nearest neighbors method; convolutional neural networks; principal component analysis; support vector machines
• Language: croatian
• URN:NBN: urn:nbn:hr:168:954560
• Study programme: Title: Computing; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva (sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Open access
• Created on: 2024-10-12 21:49:01