Sustav dubokog učenja za analizu videozapisa sportskih susreta

master's thesis

• Title (english): Deep Learning System for Analysis of Video Sequences of Sport Matches
• Author: Vito Sabalić
• Mentor: Tomislav Hrkać (mentor)
• Committee member: Tomislav Hrkać (predsjednik povjerenstva)
• Committee member: Siniša Šegvić (član povjerenstva)
• Committee member: Zoran Kalafatić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2024-07-11, Croatia

Abstract

Potražnja za naprednom analizom sportskih susreta neprestano raste, a automatski sustavi za praćenje igrača postaju ključni alati za razumijevanje dinamike sportske utakmice. Ovaj diplomski rad bavi se razvojem sustava za automatsku detekciju i praćenje igrača na teniskom terenu, koristeći ulazne videozapise teniskih utakmica. Sustav kombinira detektor YOLOv8 za detekciju igrača, algoritam DeepSORT za praćenje igrača iz kadra u kadar te matricu homografije za preslikavanje pozicija igrača na dvodimenzionalni prikaz teniskog terena. U ovom radu temeljito je objašnjena teorijska podloga navedenih metoda, opisana je implementacija tih metoda u sustav te je predstavljen pseudokod implementacije svih komponenata. Također, prikazani su rezultati izvođenja sustava te su evaluirane preciznosti detektora i preslikavanja matricom homografije. Na kraju rada istaknuta su ograničenja implementacije te su predložena potencijalna unaprjeđenja.

Abstract (english)

The demand for advanced analysis of sports events is continually increasing, and automatic player tracking systems are becoming crucial tools for understanding the dynamics of a sports game. This thesis focuses on the development of a system for automatic detection and tracking of players on a tennis court, using video recordings of tennis matches as the input. The system combines the YOLOv8 detector for player detection, the DeepSORT algorithm for tracking players frame by frame, and a homography matrix for mapping player positions onto a two-dimensional representation of the tennis court. This work thoroughly explains the theoretical foundation of the aforementioned methods, describes their implementation in the system, and presents pseudocode for all implemented components. Furthermore, the system's execution results were displayed and the accuracy of the detector and homography matrix mapping were evaluated. Finally, the limitations of the implementation were discussed along with potential improvements.

• Keywords: Sportska analiza; Detekcija igrača; YOLOv8; Praćenje igrača; DeepSORT; Matrica homografije
• Keywords (english): Sport analysis; Player detection; YOLOv8; Player tracking; DeepSORT; Homography matrix
• Language: croatian
• URN:NBN: urn:nbn:hr:168:286438
• 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: 2024-10-12 22:03:30