Detekcija i praćenje igrača u snimkama nogometnih utakmica

master's thesis

• Title (english): Player Detection and Tracking in Football Match Video
• Author: Nikola Duvnjak
• Mentor: Zoran Kalafatić (mentor)
• Committee member: Zoran Kalafatić (predsjednik povjerenstva)
• Committee member: Željka Mihajlović (član povjerenstva)
• Committee member: Tomislav Hrkać (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2022-02-24, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

U ovom radu proučeni su postojeći pristupi računalnog vida za detekciju i praćenje objekata te je implementiran sustav za praćenja igrača u snimkama nogometnih utakmica korištenjem SMOT radnog okvira. Implementirani sustav koristi shemu ponovnog praćenja kojom se naglašava istodobno izvođenje detekcije i generiranja trajektorija. Za detekciju objekata, sustav se oslanja na "single-shot" detektor objekata. Rezultati sustava vizualiziraju se i vrednuju računanjem kvantitativnih metrika. Sustav postiže zadovoljavajuće rezultate na video-snimkama niske rezolucije, no shema ponovnog praćenja, zbog istovremenog izvođenja detekcije i generiranja trajektorije, nije pogodna za spajanje dijelova detekcija na slikama visoke rezolucije.

Abstract (english)

In this thesis, the existing approaches of computer vision for object detection and tracking are studied, and a system for tracking players, in footage of football matches, is implemented using the SMOT framework. The implemented system uses a re-detection scheme which emphasizes simultaneous performance of object detection and tracklet generation. For object detection, the system relies on a "single-shot" object detector. The results of the system are visualized and evaluated by calculating quantitative metrics. The system achieves satisfactory results on low-resolution videos. However, re-detection scheme is not suitable for merging partial detection results in high-resolution image, due to the simultaneous performance of detection and tracklet generation.

• Keywords: računalni vid; praćenje više objekata; detekcija objekata; duboke konvolucijske neuronske mreže
• Keywords (english): computer vision; multiple object tracking; object detection; deep convolutional neural networks
• Language: croatian
• URN:NBN: urn:nbn:hr:168:366444
• 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: 2023-01-13 12:10:27