Generiranje GBSAR radarskih podataka primjenom uvjetnih generativnih suparničkih mreža

master's thesis

• Title (english): Generating GBSAR radar data using conditional generative adversarial networks
• Author: Grigor Škarić
• Mentor: Marko Bosiljevac (mentor)
• Committee member: Marko Bosiljevac (predsjednik povjerenstva)
• Committee member: Sonja Grgić (član povjerenstva)
• Committee member: Dario Bojanjac (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2024-07-08, Croatia

Abstract

GBSAR korišten je za snimanje objekata različitih oblika i materijala. Zbog velike količine vremena potrebne za prikupljanje dovoljnog broja kvalitetnih primjera za istraživanja koja koriste duboke modele, istražen je način umjetnog stvaranja uzoraka. Korištenjem više modela generativnih suparničkih mreža uspješno su generirane uvjerljive matrice radarskih signala. Uspoređene su performanse modela treniranih uz različite funkcije gubitka. Primjeri različitih klasa generirani su na dva načina, treniranjem zasebnog modela za svaku klasu te primjenom uvjetnih modela. Prvi način pruža bolje performanse, a drugi fleksibilnost.

Abstract (english)

GBSAR was used to record objects of different shapes and materials. Due to the large amount of time required to collect a sufficient number of quality examples for research using deep models, a method of artificially generating samples was investigated. By using multiple generative adversarial network models, plausible radar signal matrices were successfully generated. The performance of models trained with different loss functions was compared. Examples of different classes are generated in two ways, by training a separate model for each class and by applying conditional models. The first way provides better performance, and the second way provides flexibility.

• Keywords: Radar; FMCW radar; GBSAR; Duboko učenje; GAN
• Keywords (english): Radar; FMCW radar; GBSAR; Deep learning; GAN
• Language: croatian
• URN:NBN: urn:nbn:hr:168:559935
• 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 22:16:24