prikaz prve stranice dokumenta Metrička ugrađivanja za reprezentiranje osoba u slikama
No public access
master's thesis
Metrička ugrađivanja za reprezentiranje osoba u slikama
Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing, 2019. urn:nbn:hr:168:043723
Kačan, Marin
University of Zagreb
Faculty of Electrical Engineering and Computing

Institutional repository: FER Repository

Cite this document

Kačan, M. (2019). Metrička ugrađivanja za reprezentiranje osoba u slikama (Master's thesis). Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing. Retrieved from https://urn.nsk.hr/urn:nbn:hr:168:043723

Kačan, Marin. "Metrička ugrađivanja za reprezentiranje osoba u slikama." Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, 2019. https://urn.nsk.hr/urn:nbn:hr:168:043723

Kačan, Marin. "Metrička ugrađivanja za reprezentiranje osoba u slikama." Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, 2019. https://urn.nsk.hr/urn:nbn:hr:168:043723

Kačan, M. (2019). 'Metrička ugrađivanja za reprezentiranje osoba u slikama', Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, accessed 26 December 2024, https://urn.nsk.hr/urn:nbn:hr:168:043723

Kačan M. Metrička ugrađivanja za reprezentiranje osoba u slikama [Master's thesis]. Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing; 2019 [cited 2024 December 26] Available at: https://urn.nsk.hr/urn:nbn:hr:168:043723

M. Kačan, "Metrička ugrađivanja za reprezentiranje osoba u slikama", Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, Zagreb, 2019. Available at: https://urn.nsk.hr/urn:nbn:hr:168:043723

Metadata
TitleMetrička ugrađivanja za reprezentiranje osoba u slikama
Title (english)Metric embeddings for representing persons in images
AuthorMarin Kačan
MentorSiniša Šegvić (mentor)
Committee memberSiniša Šegvić (predsjednik povjerenstva)
Committee memberZoran Kalafatić (član povjerenstva)
Committee memberTomislav Hrkać (član povjerenstva)
GranterUniversity of Zagreb
Faculty of Electrical Engineering and Computing
Zagreb
Defense date and country2019-09-27, Croatia
Scientific / art field,
discipline and subdiscipline		TECHNICAL SCIENCES
Computing
Abstract
Metrička ugrađivanja vrlo su važan sastojak primjena u kojima ostvarujemo korespondenciju slika iste osobe u različitim trenutcima. Posebno su zanimljive primjene vezane uz praćenje i ponovnu identifikaciju osoba. U posljednje vrijeme, veliki uspjeh u tom području ostvaruju duboki modeli učeni različitim metričkim gubitcima. U okviru rada, proučeni su i ukratko opisani postojeći pristupe za učenje metričkih ugrađivanja. U okviru rada, uhodani su postupci nadziranog učenja prikladnih dubokih modela. Napravljena je usporedba utjecaja različitih metričkih ugrađivanja i različitih postupaka detekcije objekata na performanse postupka praćenja. Prikazani su rezultati eksperimenata i predloženi su pravci budućeg razvoja.
Abstract (english)
Metric embeddings are very important when we need to achieve a correspondence between images of the same person at different moments in time. Applications in multiple person tracking and person re-identification are particularly interesting. As of late, deep models trained with different metric losses have been very successful in that area. In this thesis, we use supervised learning to train different deep models for this task. We also compare the effects of different metric embeddings and different object detection procedures on the performance of the tracking procedure. We show the experiment results and suggest promising directions for future work.
Keywords
Keywords (english)
Languagecroatian
URN:NBNurn:nbn:hr:168:043723
Study programmeTitle: Computing
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 conditionsClosed access
Terms of useLicence In copyright icon
Public note
Created on2020-01-07 18:55:06