Raspoznavanje lica

master's thesis

• Title (english): Face Recognition
• Author: Neven Pičuljan
• Mentor: Igor Sunday Pandžić (mentor)
• Committee member: Igor Sunday Pandžić (predsjednik povjerenstva)
• Committee member: Mirko Sužnjević (član povjerenstva)
• Committee member: Sanja Grubeša (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2016-07-07, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Opisan je razvoj sustava za automatsko raspoznavanje lica. Automatsko raspoznavanje lica zasniva se na tehnikama računalnog vida, a služi za raspoznavanje identiteta čovjeka na osnovu slike lica, pri čemu se koristi usporedba s ranije registriranim slikama lica za koje je poznat identitet. Skup ranije registriranih lica naziva se galerija. Korištene su duboke konvolucijske neuronske mreže. Duboke konvolucijske neuronske mreže sažimaju slike u 128-dimenzionalni deskriptor, a Euklidske udaljenosti deskriptora lica izravno odgovaraju sličnostima lica.

Abstract (english)

This paper describes the development of a system for automatic face recognition. Automatic face recognition is based on the techniques for computer vision and is used to identify the identity of the person based on the facial image. It uses a comparison with previously registered facial images with identities that are known. A set of previously registered faces is called the gallery. Deep convolutional neural network is used for face recognition. It compresses the image into 128-dimensional descriptor. Euclidean distance between descriptors directly corresponds to the similarities between the faces.

• Keywords: raspoznavanje lica; duboka konvolucijska neuronska mreža; strojno učenje; triplet loss; AdaDelta; algoritam unazadne propagacije; stohastički gradijentni spust; Torch framework
• Keywords (english): face recognition; deep convolutional neural network; machine learning; triplet loss; AdaDelta; backpropagation algorithm; stochastic gradient descent; Torch framework
• Language: croatian
• URN:NBN: urn:nbn:hr:168:592383
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2018-02-06 11:17:49