20.500.12556/DKUM-75089 Razvoj naprednega sistema za detektiranje voznih pasov na platformah GPU master thesis Development of an advanced system for lane detection on GPU platforms Problem, ki ga obravnavamo v magistrski nalogi je detektiranje voznih pasov na RGB slikah oz. posnetkih ceste pred vozilom med vožnjo. Za rešitev tega problema smo se odločili uporabiti tehnike »globokega učenja«, predvsem konvolucijske nevronske mreže, s katerimi smo izvedli semantično segmentiranje. Problem smo reševali s tremi različnimi arhitekturami nevronskih mrež, ki smo jih učili na naboru podatkov BDD100k. Modele mrež smo nato testirali in primerjali rezultate s pomočjo IoU metrike za semantično segmentacijo. Opravili smo tudi več eksperimentov s ciljem izboljšanja IoU vrednosti in generalizacije modelov. Na koncu smo modele testirali tudi na Nvidia Jetson TX2 platformi in predlagali možnost vključitve takšnih modelov v sistem avtonomnega vozila. The problem we are dealing with in this master’s thesis is lane detection on RGB images, i.e. images of the road in front of the vehicle during driving. To solve this problem, we have used “deep learning” techniques, specifically convolutional neural networks for the semantic segmentation task. We designed three different network architectures, which were trained on BDD100k dataset. Those network models were then evaluated and compared based on IoU metric used for semantic segmentation. We performed several experiments with the goal of improving IoU results and the generalization of the models. Finally, we tested the models on the Nvidia Jetson TX2 platform and proposed the possibilities for incorporating such models into the autonomous vehicle system. globoko učenje konvolucijske nevronske mreže segmentacija voznega pasu strojni vid avtonomno vozilo deep learning convolutional neural networks lane segmentation computer vision autonomous vehicles true true false [K. Crnek] Slovenski jezik Angleški jezik Magistrsko delo/naloga 2019-09-20 10:43:23 2019-11-13 09:09:35 2022-08-09 09:57:05 2019-12-10 08:19:49 2019 0 Maribor 0 Maribor XVII, 84 f. 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 621.396.969.3:004.89(043.2) 22831638 URN:SI:UM:DK:HACS1X2M MAG_Crnek_Karlo_2019.pdf MAG_Crnek_Karlo_2019.pdf 1 3BD144AC551685CAE9C9112C206E3470 ab88ad229e1345f57671e27c7b4fa6e1872e3a1936d8c955da0ad85a2dcd3e96 021cc763-7c11-11eb-bb7a-00155d0001ca 20.500.12556/dkum/e84621ae-b288-4448-ad60-2723a111ce40 https://dk.um.si/Dokument.php?lang=slv&id=138325 Fakulteta za elektrotehniko, računalništvo in informatiko 0 0 0