20.500.12556/DKUM-84091 Primerjava zaznave in segmentacije neznanih objektov v 6 prostostnih stopnjah magistrsko delo Comparison of detection and segmentation of unknown objects for 6 dof pose estimation Cilj diplomskega dela je analizirati različne pristope odkrivanja in segmentacije neznanih predmetov ter primerjati podobnosti in razlike algoritmov. Raziskava temelji na ‟No Free Lunch Theorem‟ in je osredotočena na iskanje najustreznejših pristopov strojnega vida v robotiki in primerjavo njihove učinkovitosti. Na začetku je narejen pregled stanja raziskav z delitvijo na regresijske in klasifikacijske metode, zatem sledi segmentacija in prepoznava objektov ter se zaključi z opisom klasičnih in metod z nevronskim mreženjem. V nadaljevanju je opredeljena primerjava nabora podatkov in že narejenih meritev. Zadnji dve poglavji sta namenjeni evalvaciji lastnih poskusov in zaključek s priložnostmi nadaljnjih raziskav na tem področju. The thesis work aims to analyse different approaches of detection and segmentation of unknown objects and to compare the similarities and differences of algorithms. Research is based on No Free Lunch Theorem on finding the most appropriate approaches of machine vision in robotics and comparing their efficiency. At the beginning, an overview of the state of the art is made, dividing it into regression and classification methods, followed by segmentation and object recognition, and ends with a description of classical and neural network methods. In the following, a comparison of the data set and already made measurements is defined. The last two chapters are devoted to the evaluation of own experiments and the conclusion with opportunities for further research in this area. 6-DoF strojni vid evalvacija pozicije segmentacija slike zaznavanje objektov 6-DoF machine vision pose evaluation image segmentation object detection true true false [N. Herženjak] Slovenski jezik Angleški jezik Magistrsko delo/naloga 2023-04-10 20:46:45 2023-06-08 07:55:16 2023-11-08 14:00:02 0000-00-00 00:00:00 2023 0 Maribor 0 Maribor 1 spletni vir (1 datoteka PDF (VII, 77 f.)) 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 004.932(043.2) 161956355 MAG_Herzenjak_Nejc_2023.pdf MAG_Herzenjak_Nejc_2023.pdf 1 3740C8ADF51B9EAFFD9F7043039D74D5 b1ffc7ede58bb7fb7bb71b60135d77820242855152031e7be1abbd8f203bad6a c418fa45-0309-11ee-b87f-00155d000105 https://dk.um.si/Dokument.php?lang=slv&id=167741 Fakulteta za strojništvo 0 0 0