Poboljšanje rezultata detekcije kroz variranje izgleda objekta

undergraduate thesis

• Title (english): Improving of Detection Results Through Variation of Object Appearance
• Author: Fran Hrabar
• Mentor: Marko Subašić (mentor)
• Committee member: Marko Subašić (predsjednik povjerenstva)
• Committee member: Denis Milošević (član povjerenstva)
• Committee member: Karlo Koščević (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2021-07-07, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Tema ovog rada je poboljšanje rezultata procjene dubinske udaljenosti objekta od kamere kroz variranje teksture 3D objekta. U svrhu uključenja iscrtavanja slika i promjene teksture objekta u cjevovod koristi se diferencijabilni prikazivač PyTorch3D. Detaljno su opisani parametri scene prikazivača i skup podataka. Dotaknuta je i tema konvolucijskih neuronskih mreža i njihova primjena u detekciji objekata kao i EfficientNet u ulozi ekstraktora značajki. Na kraju rada prikazani su rezultati testiranja naučenih modela sa promjenama tekstura na skupu za testiranje.

Abstract (english)

The topic of this paper is to improve the results of depth estimation of an object from camera through the variation of the texture of a 3D object. A differentiable PyTorch3D renderer is used to include image rendering and variation of the texture of a object in the pipeline. The renderer’s scene parameters and data set are described in detail. The topic of convolutional neural networks and their application in object detection as well as EfficientNet in feature extractor role were also touched upon. At the end of the paper, the results of testing the learned models with the change of texture on the testing set are presented.

• Keywords: duboko učenje; računalni vid; umjetne neuronske mreže; duboke neuronske mreže; konvolucijske neuronske mreže; diferencijabilni prikazivač; regresija; procjena dubine
• Keywords (english): deep learning; computer vision; artificial neural network; deep neural networks; convolutional neural networks; differentiable renderer; regression; depth estimation
• Language: croatian
• URN:NBN: urn:nbn:hr:168:627479
• 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: 2023-01-14 20:35:11