Prepoznava obrazov z opisnikom Alfa-Gama

Lušina, Domen (author), Kristan, Matej (mentor), Maver, Jasna (comentor)

Za prepoznavo obrazov obstajajo številne metode, nekatere med njimi za prepoznavo uporabljajo lokalne opisnike. V tej diplomskem delu si bomo za namen prepoznave obrazov ogledali delovanje popolnoma novega lokalnega opisnika, imenovanega Alfa-Gama (krajše AG), ki je bil razvit v Laboratoriju za vizualne in spoznavne sisteme na Fakulteti za računalništvo in informatiko v Ljubljani. Pripravili smo sistem za prepoznavo obrazov na podlagi algoritma vreče besed in klasifikatorja SVM . Predlagan sistem smo razvili za številne popularne opisnike in opisnik AG. Sistem smo testirali na javno dostopnih podatkovnih bazah. V večini primerov je opisnik AG izboljšal rezultate prepoznave in se je izkazal za dobro alternativo k ostalim testiranim opisnikom.

• Language: Slovenian
• Keywords: Alfa-Gama opisnik, računalniški vid, prepoznava obraza, detekcija obraza, lokalni opisniki
• Work type: Bachelor thesis/paper
• Organization: FRI - Faculty of Computer and Information Science
• Year: 2017
• PID: 20.500.12556/RUL-95926
• Publication date in RUL: 25.09.2017

LUŠINA, Domen, 2017, Prepoznava obrazov z opisnikom Alfa-Gama [online]. Bachelor’s thesis. [Accessed 29 March 2025]. Retrieved from: https://repozitorij.uni-lj.si/IzpisGradiva.php?lang=eng&id=95926

Secondary language

• Language: English
• Title: Face recognition using Alapha-Gamma descriptor

Abstract

For facial recognition there are numerous methods available, some of them use local descriptors. In this thesis, for the purpose of face recognition, we will take a closer look at novel local descriptor Alpha-Gamma (shorter AG) developed at the Visual Cognitive Systems Laboratory at the Faculty Of Computer and Information Science, University of Ljubljana. We developed a system for face recognition using bag of words algorithm and SVM classifier. The proposed system was developed for several popular local descriptors and AG. We used publicly available databases for testing. In most cases AG improved facial recognition rate and was shown as a good alternative to other tested local descriptors.

• Keywords: Alpha-Gamma descriptor, computer vision, facial recognition, face detection, local descriptors