prikaz prve stranice dokumenta Building a labeled dataset for recognition of handball actions using mask R-CNN and STIPS
  Preuzmi
PDF 935.05 KB
Znanstveni rad - Izvorni znanstveni rad
Building a labeled dataset for recognition of handball actions using mask R-CNN and STIPS
U: 2018 7th European Workshop on Visual Information Processing (EUVIP). IEEE, 2019. str. 1-6. https://doi.org/10.1109/EUVIP.2018.8611642
Ivašić-Kos, Marina; Pobar, Miran

Institucijski repozitorij: Repozitorij Fakulteta informatike i digitalnih tehnologija Sveučilišta u Rijeci

Citirajte ovaj rad

Ivašić Kos, M. i Pobar, M. (2019). Building a labeled dataset for recognition of handball actions using mask R-CNN and STIPS. 2018 7th European Workshop on Visual Information Processing (EUVIP) (str. 1-6). IEEE. doi: 10.1109/EUVIP.2018.8611642

Ivašić Kos, Marina i Miran Pobar. "Building a labeled dataset for recognition of handball actions using mask R-CNN and STIPS." 2018 7th European Workshop on Visual Information Processing (EUVIP), IEEE, 2019, str. 1-6. https://urn.nsk.hr/urn:nbn:hr:195:898413

Ivašić Kos, Marina i Miran Pobar. "Building a labeled dataset for recognition of handball actions using mask R-CNN and STIPS." U 2018 7th European Workshop on Visual Information Processing (EUVIP), 1-6. IEEE, 2019. Citirano 27.11.2024.. https://urn.nsk.hr/urn:nbn:hr:195:898413

Ivašić Kos, M. i Pobar, M. (2019) 'Building a labeled dataset for recognition of handball actions using mask R-CNN and STIPS', 2018 7th European Workshop on Visual Information Processing (EUVIP), IEEE, str. 1-6. Preuzeto s: https://urn.nsk.hr/urn:nbn:hr:195:898413 (Datum pristupa: 27.11.2024.)

Ivašić Kos M, Pobar M. Building a labeled dataset for recognition of handball actions using mask R-CNN and STIPS. 2018 7th European Workshop on Visual Information Processing (EUVIP). IEEE; 2019. Str. 1-6. [pristupljeno 27.11.2024.] Dostupno na: https://urn.nsk.hr/urn:nbn:hr:195:898413

M. Ivašić Kos i M. Pobar, "Building a labeled dataset for recognition of handball actions using mask R-CNN and STIPS", 2018 7th European Workshop on Visual Information Processing (EUVIP) IEEE, 2019. [Online] Dostupno na: https://urn.nsk.hr/urn:nbn:hr:195:898413 [Citirano: 27.11.2024.]

Za citiranje koristite ovu mrežnu adresu: https://urn.nsk.hr/urn:nbn:hr:195:898413
Podaci o radu
Naslov (engleski)Building a labeled dataset for recognition of handball actions using mask R-CNN and STIPS
AutorMarina Ivašić-Kos https://orcid.org/0000-0002-1940-5089
AutorMiran Pobar https://orcid.org/0000-0001-5604-2128
Autorova ustanovaSveučilište u Rijeci
(Fakultet informatike i digitalnih tehnologija)
Znanstveno / umjetničko
područje, polje i grana		DRUŠTVENE ZNANOSTI
Informacijske i komunikacijske znanosti
Znanstveno / umjetničko
područje, polje i grana		TEHNIČKE ZNANOSTI
Računarstvo
Umjetna inteligencija
Sažetak (engleski)
Building successful machine learning models depends on large amounts of training data that often needs to be labelled manually. We propose a method to efficiently build an action recognition dataset in the handball domain, focusing on minimizing the manual labor required to label the individual players performing the chosen actions. The method uses existing deep learning object recognition methods for player detection and combines the obtained location information with a player activity measure based on spatio-temporal interest points to track players that are performing the currently relevant action, here called active players. The method was successfully used on a challenging dataset of real-world handball practice videos, where the leading active player was correctly tracked and labeled in 84 % of cases.
Ključne riječi (engleski)
Jezikengleski
Vrsta publikacijeZnanstveni rad - Izvorni znanstveni rad
Status objaveObjavljen
Vrsta recenzijeRecenziran - međunarodna recenzija
Verzija publikacijeZavršna verzija rukopisa prihvaćena za objavljivanje (postprint)
Naslov zbornika2018 7th European Workshop on Visual Information Processing (EUVIP)
Brojčani podacistr. 1-6
ISBN978-1-5386-6897-9 (Electronic ISBN)
ISBN978-1-5386-6898-6 (Print on Demand(PoD) ISBN)
e-ISSN2471-8963
DOIhttps://doi.org/10.1109/EUVIP.2018.8611642
URN:NBNurn:nbn:hr:195:898413
Datum objave publikacije2019-01-17
ProjektŠifra: IP-2016-06-8345
Naziv (hrvatski): Automatsko raspoznavanje akcija i aktivnosti u multimedijalnom sadržaju iz domene sporta
Naziv (engleski): Automatic recognition of actions and activities in multimedia content from sports domain
Kratica: RAASS
Voditelj: Marina Ivašić Kos
Pravna nadležnost: Hrvatska
Financijer: HRZZ
Linija financiranja: IP
SkupNaziv: 2018 7th European Workshop on Visual Information Processing (EUVIP)
Kratica: EUVIP
Lokacija: Tampere, Finska
Datum početka: 2018-11-26
Datum završetka: 2018-11-28
Vrsta izlaganja: Predavanje
URL dokumentahttps://ieeexplore.ieee.org/document/8611642
Vrsta resursaTekst
IzdavačIEEE
Prava pristupaOtvoreni pristup
Uvjeti korištenjaIkona licence Zaštićeno autorskim pravom.
Javna napomena© 2018 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.
Datum i vrijeme pohrane2023-02-12 19:44:39