prikaz prve stranice dokumenta Podudaranje osoba na osnovu histograma boje
Access restricted to students and staff of home institution
master's thesis
Podudaranje osoba na osnovu histograma boje
2018. urn:nbn:hr:190:731923
Pinčić, Domagoj
University of Rijeka
Faculty of Engineering

Cite this document

Pinčić, D. (2018). Podudaranje osoba na osnovu histograma boje (Master's thesis). Rijeka: University of Rijeka, Faculty of Engineering. Retrieved from https://urn.nsk.hr/urn:nbn:hr:190:731923

Pinčić, Domagoj. "Podudaranje osoba na osnovu histograma boje." Master's thesis, University of Rijeka, Faculty of Engineering, 2018. https://urn.nsk.hr/urn:nbn:hr:190:731923

Pinčić, Domagoj. "Podudaranje osoba na osnovu histograma boje." Master's thesis, University of Rijeka, Faculty of Engineering, 2018. https://urn.nsk.hr/urn:nbn:hr:190:731923

Pinčić, D. (2018). 'Podudaranje osoba na osnovu histograma boje', Master's thesis, University of Rijeka, Faculty of Engineering, accessed 15 August 2024, https://urn.nsk.hr/urn:nbn:hr:190:731923

Pinčić D. Podudaranje osoba na osnovu histograma boje [Master's thesis]. Rijeka: University of Rijeka, Faculty of Engineering; 2018 [cited 2024 August 15] Available at: https://urn.nsk.hr/urn:nbn:hr:190:731923

D. Pinčić, "Podudaranje osoba na osnovu histograma boje", Master's thesis, University of Rijeka, Faculty of Engineering, Rijeka, 2018. Available at: https://urn.nsk.hr/urn:nbn:hr:190:731923

Please login to the repository to save this object to your list.
Metadata
TitlePodudaranje osoba na osnovu histograma boje
AuthorDomagoj Pinčić
MentorKristijan Lenac (mentor)
Committee memberKristijan Lenac (predsjednik povjerenstva)
GranterUniversity of Rijeka
Faculty of Engineering
Rijeka
Defense date and country2018-05, Croatia
Scientific / art field,
discipline and subdiscipline		TECHNICAL SCIENCES
Computing
Abstract
Rezultati dobiveni u ovom diplomskom radu pokazuju kako metode reidentifikacije osoba temeljene na histogramu boja pokazuju obećavajuće rezultate u problematici reidentifikacije osoba u RGB videozapisima. Prethodno samom procesu reidentifikacije, na svakoj slici unutar videozapisa su detektirane osobe koristeći MobileNetSSD neuralnu mrežu za detekciju objekata. Nakon toga se osobama pridjeljuje jedinstveni identifikacijski broj i osoba se prati dok god se nalazi u kadru kamere. Po izlasku iz kadra izvršava se proces reidentifikacije osobe odnosno usporedbe osobe sa svim osobama u bazi podataka. Uz procese detekcije, praćenja i reidentifikacije osoba izvršava se i segmentacija osobe od pozadine što pridonosi točnosti metoda reidentifikacije. Također, prikazana je usporedba točnosti nekoliko metoda reidentifikacije kao i testiranje metode bazirane na 3D histogramu boja u HSV modelu boja na snimljenom setu podataka koji uključuje 21 različitu osobu.
Abstract (english)
Results obtained in this work shows that methods for person re-identification based on color histograms show promising results when used on re-identifying persons in RGB video records. Before re-identification process persons are firstly detected in video record frames using MobileNetSSD neural network for object detection. After detection each person is assigned a unique identifying number and tracked as long as person is in the camera frame. After person has left the frame re-identification process is executed, where person is compared with all other persons in database. Along person detection, tracking and re-identification segmentation of person from background is implemented, which improves accuracy of re-identification methods. Also, accuracy comparison of several re-identification methods is shown, with additional tests on method based on 3D HSV color histogram on custom dataset made of 21 different persons.
Keywords
Keywords (english)
Languagecroatian
URN:NBNurn:nbn:hr:190:731923
Study programmeTitle: Computer Engineering
Study programme type: university
Study level: graduate
Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
Type of resourceText
File originBorn digital
Access conditionsAccess restricted to students and staff of home institution
Terms of useLicence In copyright icon
Created on2018-06-05 09:16:34