diplomsko delo
Grega Potočnik (Author), Matej Kristan (Mentor), Alan Lukežič (Co-mentor)

Abstract

Diplomsko delo obravnava izzive sledenja posameznemu objektu v videoposnetkih s prisotnostjo motilcev. Glavni cilj je razviti metodo, ki z uporabo grafovnih nevronskih mrež in dodatnih informacij o okoliških objektih izboljša stabilnost in natančnost sledilnika. Predlagamo metodo nevronskega povezovalnega algoritma z gručenjem (angl. Neural Solver with Grouping, NSG), ki temelji na štiristopenjskem cevovodu: sledenje z uporabo nekega sledilnika za osnovno detekcijo, generator hipotez, ki generira dodatne kandidate za lokacije objekta, nevronski povezovalni algoritem za povezovanje detekcij v trajektorije ter nazadnje še gručenje sledi s Kalmanovim filtrom za napovedovanje pozicij in ponovno identifikacijo ob izgubi sledi. Eksperimentalno delo je vključevalo evaluacijo na podatkovnih zbirkah LaSOT in DiDi z metrikami orodja VOT. Rezultati so pokazali izboljšano kakovost sledenja za 1.2 odstotka v primerjavi z uporabljenim sledilnikom za osnovno detekcijo. Omejitve metode so povezane z občutljivostjo na kakovost vhodnih detekcij in težave pri zelo dinamičnih scenarijih. Nadaljnji razvoj bi lahko vključeval integracijo algoritmov za gručenje v globoke nevronske mreže ter optimizacijo generatorja hipotez. Predlagana rešitev predstavlja korak k robustnejšim sistemom za sledenje v realnem času.

Keywords

računalniški vid;nevronske mreže;sledenje objektom;videoposnetki;motilci;univerzitetni študij;diplomske naloge;

Data

Language: Slovenian
Year of publishing:
Typology: 2.11 - Undergraduate Thesis
Organization: UL FRI - Faculty of Computer and Information Science
Publisher: [G. Potočnik]
UDC: 004.93(043.2)
COBISS: 241542915 Link will open in a new window
Views: 126
Downloads: 40
Average score: 0 (0 votes)
Metadata: JSON JSON-RDF JSON-LD TURTLE N-TRIPLES XML RDFA MICRODATA DC-XML DC-RDF RDF

Other data

Secondary language: English
Secondary title: Object tracking in presence of distractors using graph neural networks
Secondary abstract: The thesis addresses the challenges of single-object tracking in videos in the presence of distractors. The main objective is to develop a method that leverages graph neural networks and additional information about surrounding objects to improve the stability and accuracy of the tracker. The proposed method, NSG (Neural Solver with Grouping), is based on a four-stage pipeline: initial tracking using a baseline tracker for primary detections, a hypothesis generator that produces additional candidate object locations, a neural solver for associating detections into trajectories, and finally, trajectory grouping with a Kalman filter for position prediction and re-identification in case of track loss. The experimental evaluation was conducted on the LaSOT and DiDi datasets using metrics from the VOT toolkit. Results showed an improvement in tracking quality by 1.2 percentage points compared to the baseline tracker. The method’s limitations are primarily related to sensitivity to input detection quality and challenges in highly dynamic scenarios. Future work could include integrating clustering algorithms into deep neural networks and optimizing the hypothesis generator. The proposed solution represents a step towards more robust real-time tracking systems.
Secondary keywords: computer vision;neural networks;object tracking;videos;distractors;computer science;diploma;
Type (COBISS): Bachelor thesis/paper
Study programme: 1000468
Thesis comment: Univ. v Ljubljani, Fak. za računalništvo in informatiko
Pages: 1 spletni vir (1 datoteka PDF (57 str.))
ID: 26719402