Praćenje igrača u snimkama nogometnih utakmica

master's thesis

• Title (english): Tracking of Players in Football Game Video
• Author: Matija Oršolić
• Mentor: Zoran Kalafatić (mentor)
• Committee member: Zoran Kalafatić (predsjednik povjerenstva)
• Committee member: Tomislav Hrkać (član povjerenstva)
• Committee member: Siniša Šegvić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2018-07-17, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Proučene su mogućnosti naknadne obrade rezultata algoritma za praćenje više objekata. Opisani su i međusobno uspoređeni sljedeći algoritmi za praćenje objekata: Boosting, MIL, KCF, Median Flow i TLD. Usporedba je izvršena nad pet minutnom video sekvencom nogometne utakmice, koja je zbog kompleksnosti problema podijeljena u više malih video sekvenci od kojih svaka testira specifični slučaj. Opisani su neki načini evaluacije algoritama za praćenje objekata. Za evaluaciju i usporedbu korištena je CLEAR MOT metoda koja kao izlaz daje preciznost i točnost algoritma za praćenje više objekata. Konačno, prikazani su rezultati evaluacije i usporedbe te je na temelju rezultata zaključeno da se uz pomoć razmatranih algoritama za praćenje objekta naknadna ljudska obrada izlaza algoritma za praćenje više objekata ne može smanjiti.

Abstract (english)

Possibilities for post processing of multi-object tracking algorithm were studied. The following object tracking algorithms have been described and compared: Boosting, MIL, KCF, Median Flow and TLD. The comparison was made over a five-minute football game video sequence, divided into several small sequences because of the complexity of the problem, each of which tests a specific case. Some methods of evaluating object tracking algorithms were described. Evaluation and comparison was done using the CLEAR MOT metrics, which describes the precision and accuracy of the multiple object tracking algorithm. Finally, the results of the evaluation and comparison were presented, and it was concluded that with the help of considered trackers, manual post processing of multiple object tracker output could not be reduced.

• Keywords: Računalni vid; praćenje objekata; evaluacija algoritama za praćenje; naknadna obrada rezultata praćenja; Boosting; MIL; KCF; Median Flow; TLD; CLEAR MOT
• Keywords (english): Computer vision; object tracking; tracking algorithm evaluation; tracking results post processing; Boosting; MIL; KCF; Median Flow; TLD; CLEAR MOT
• Language: croatian
• URN:NBN: urn:nbn:hr:168:071406
• 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: 2019-03-13 23:32:30