magistrsko delo
Aleks Vujić (Author), Vlado Stankovski (Mentor)

Abstract

Povpraševanje po najnovejših modelih umetne inteligence za analizo strogo varovanih videotokov, ki jih na enostaven način namestimo v oblak, je vse večje. Poraja se potreba po zaupanja vrednem tržnem ekosistemu, ki omogoča varno interakcijo s ponudniki virov. Z uporabo decentraliziranih aplikacij lahko dosežemo transparentnost delovanja in večje zaupanje uporabnikov v sistem. V okviru magistrske naloge smo predstavili zaledne, čelne in orkestracijske tehnologije, ki jih potrebujemo za razvoj decentraliziranih aplikacij in predlagali načrt arhitekture, ki združuje številne komponente. Opisali smo interaktivno spletno aplikacijo za analizo videotokov, ki uporabniku omogoča zagon izbrane metode umetne inteligence za analizo lastnega videotoka. Ključni podatki so shranjeni na pametni pogodbi, podatke iz zunanjih sistemov pa pridobi s prerokom. Kubernetes namesti metode umetne inteligence na infrastrukturo v obliki vsebnikov. V primerjavi s sorodnimi deli se odlikuje po monetizaciji storitev, zaupanju v sistem, dostopnosti metod in možnosti računanja na robu, kar predstavlja napredek najnovejših dosežkov na tem področju. Aplikacija, razvita v tej magistrski nalogi, lahko služi kot primer dobre prakse za nadaljnje raziskovalno delo na področju razvoja decentraliziranih aplikacij. Prikaže, kako lahko integriramo verigo blokov, orkestracijo vsebnikov, umetno inteligenco in spletne uporabniške vmesnike.

Keywords

videotok;pametna pogodba;DevOps;magisteriji;

Data

Language: Slovenian
Year of publishing:
Typology: 2.09 - Master's Thesis
Organization: UL FRI - Faculty of Computer and Information Science
Publisher: [A. Vujić]
UDC: 004.8(043.2)
COBISS: 100507395 Link will open in a new window
Views: 219
Downloads: 101
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: Decentralised System for Video Stream Analysis
Secondary abstract: The demand for the latest AI models for the analysis of strictly protected video streams, which are easily deployed to the cloud, is growing. There is a need for a trusted market ecosystem that allows for secure interaction with resource providers. By using decentralized applications, we can achieve operational transparency and increase user trust in the system. In this master's thesis, we presented backend, frontend and orchestration technologies needed to develop decentralized applications and proposed an architecture that integrates many components. We described an interactive web application for video stream analysis that allows the user to run a selected AI model to analyze their own video stream. Key data is stored on the smart contract, and data from external systems is obtained from the oracle. Kubernetes deploys AI models on infrastructure in the form of containers. Compared to related works, it is distinguished by the monetization of services, trust in the system, accessibility of methods and the possibility of edge computing, which represents the advancement of the state of the art in the field. The application developed in this master's thesis can serve as an example of good practice for further research work in the field of decentralized application development. It demonstrates how we can integrate blockchain, container orchestration, AI and web user interfaces.
Secondary keywords: blockchain;videostream;smart contract;DevOps;deep learning;computer science;computer and information science;master's degree;Verige blokov (zbirke podatkov);Globoko učenje (strojno učenje);Računalništvo;Univerzitetna in visokošolska dela;
Type (COBISS): Master's thesis/paper
Study programme: 1000471
Embargo end date (OpenAIRE): 1970-01-01
Thesis comment: Univ. v Ljubljani, Fak. za računalništvo in informatiko
Pages: 112 str.
ID: 14641672