Identifikacija osobe temeljena na TensorFlow biblioteci za strojno učenje

master's thesis

• Title (english): Person identification based on the TensorFlow machine learning library
• Author: Juraj Bračun
• Mentor: Tomislav Stipančić (mentor)
• Granter: University of Zagreb; Faculty of Mechanical Engineering and Naval Architecture; Zagreb
• Defense date and country: 2022-07-18, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Mechanical Engineering

Abstract

Područje umjetne inteligencije u današnje se vrijeme koristi gotovo u svakoj grani tehnologije
kao i u sve većem broju strojeva i predmeta kojima svaki dan rukujemo. Neki su se dijelovi
umjetne inteligenciju razvili do stupnja gdje je čovjek postao gotovo ovisan o rezultatima koje
takvi, pametni sustavi daju. Primjena takvih sustava uvelike je promijenila način na koji ljudi
pristupaju određenim tehničkim problemima. Njihova je glavna prednost, mogućnost
samostalnog i kontinuiranog učenja koje konstantno unaprjeđuje preciznost i točnost takvih
sustava, čime se oponaša djelovanje ljudskog mozga koji na isti način funkcionira u realnom
okruženju. Jedna vrsta takvih pametnih sustava koja je postala veoma popularna u zadnjih
desetak ili više godina, svakako je sposobnost računala da prepozna različita ljudska lica. Iako
je taj proces u ljudskom mozgu gotovo automatiziran te ne predstavlja nikakav poseban
napor, za računalo to je ozbiljan izazov kojega nije bilo lako riješiti.
U ovome će se radu kreirati upravo jedan program čiji će zadatak biti prepoznavanje ljudskih
lica u realnom vremenu. Da bi se to moglo ostvariti, potrebno je koristiti dva odvojena
sustava, jedan zadužen za detekciju lica na slici, a drugi zadužen za prepoznavanje osobe čije
se lice nalazi na slici. Korištenje oba sustava temelji se na tzv. prenesenom učenju, gdje se
već kreirani sustavi prilagođavaju i modificiraju za neke specifične potrebe, u ovom slučaju
za izradu sustava za prepoznavanje lica.

Abstract (english)

Today, artificial intelligence is being used in almost every aspect of technology, as well as in
growing number of machines and gadgets that we use on daily basis. Some areas of artificial
intelligence are so advanced, that people almost depend on the results which such smart
systems give on their output. Application of those systems has greatly changed the way we
approach different technical problems. Main advantage of modern, smart systems is the
capability to continuously learn by themselves, which results in much higher accuracy and
precision. This learning is based on imitating human brain that works the same way in real
environment. One example of these smart systems is definitely computer ability to recognize
human faces, which has become very popular in the last 10 or so years. Although, the process
of recognizing faces is almost automated in human brain, it has been a real challenge to
implement such thing in computers and machines.
In this paper, one such facial recognition system will be created. The main task of that system
will be recognizing faces in real time using web camera. To accomplish that, it is necessary to
use two separate subsystems, one for the face detection and the other for face recognition
once face has been detected. Both subsystems are based on transfer learning, which means
that they are already created and only need to be slightly modified for this special case of
recognizing faces in real time.

• Keywords: umjetna inteligencija; računalni vid; neuronske mreže; detekcija lica; prepoznavanje lica
• Keywords (english): artificial intelligence; computer vision; neural networks; face detection; face recognition
• Language: croatian
• URN:NBN: urn:nbn:hr:235:060838
• Study programme: Title: Mechanical Engineering; specializations in: Design, Process and Energy Engineering, Production Engineering, Engineering Modelling and Computer Simulation, Marine Engineering, Industrial Engineering and Management, Materials Engineering, Mechatronics and Robotics, Autonomous Systems; Course: Autonomous Systems; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka strojarstva (magistar/magistra inženjer/inženjerka strojarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Open access
• Created on: 2022-07-12 13:53:46