Praćenje objekata u videozapisima

undergraduate thesis

• Title (english): Object Tracking in Video Sequences
• Author: Ana Gršković
• Mentor: Tomislav Hrkać (mentor)
• Committee member: Tomislav Hrkać (predsjednik povjerenstva)
• Committee member: Zorana Doždor (član povjerenstva)
• Committee member: Franjo Matković (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2022-07-04, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Praćenje pokretnih objekata je veoma relevantan problem u području računalnog vida te je sastavna komponenta brojnih drugih algoritama umjetne inteligencije. Danas postoji mnogo algoritama koji na razne načine optimiraju praćenje pokretnih objekata, a jedan od tih algoritama je Median Flow algoritam. Median Flow algoritam je algoritam koji koristi funkciju za račun optičkog toka (rijetkog ili gustog) te uspoređuje rezultate optičkog toka izračunatog u oba smjera (unaprijed i unazad kroz vrijeme) kako bi filtrirao piksele s najtočnije izračunatim optičkim tokom i iskoristio ih za predviđanje buduće lokacije graničnog okvira objekta. U sklopu ovog rada napravljena je implementacija Median Flow algoritma koja koristi Gunnar-Farneback algoritam za račun gustog optičkog toka. Ta je implementacija pokazala dobre rezultate, mjerljive s implementacijom Median Flow algoritma u OpenCV biblioteci.

Abstract (english)

Object tracking in video sequences is a very relevant problem in the field of computer vision, and is a constituent component of many other artificial intelligence algorithms. Today, there are plenty of algorithms that try to optimize moving objekt tracking in various ways. One such algorithm is Median Flow algorithm. Median Flow is an algorithm which uses a function for calculating optical flow (sparse or dense) and compares the results of the optical flow calculated in both directions (forward and backward in time). By doing so, Median Flow algorithm manages to filter out the pixels of a video with the most accurate optical flow and use them to predict the next location of the object's bounding box. As a part of this paper, an implementation of the Median Flow algorithm is created. This implementation uses the Gunnar-Farneback algorithm for calculating dense optical flow and has shown good results, almost as good as the implementation in the OpenCV library.

• Keywords: praćenje objekata; optički tok; granični okvir; Median Flow; pogreška naprijed natrag
• Keywords (english): object tracking; optical flow; bounding box; Median Flow; forward backward error
• Language: croatian
• URN:NBN: urn:nbn:hr:168:853734
• Study programme: Title: Computing; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva (sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-01-15 10:15:14