Razvoj programske opreme za generiranje učnih množic iz CAD modelov

Industrija tem bolj stremi k temu, da roboti opravljajo vse naloge. Glaven problem pa je, da se roboti ne znajo programirat sami in prepoznavati predmetov in okolice brez kakršnega koli treniranja. Tu nastopi ta zaključna naloga, katere namen je razvoj programske opreme za ustvarjanje slik iz CAD mo...

Full description

Bibliographic Details
Main Author: Šketa, Tilen
Other Authors: Bračun, Drago
Format: Bachelor Thesis
Language:Slovenian
Published: T. Šketa 2023
Subjects:
program za generiranje slik
CAD model
podatkovne baze
Python program
Blender
image generation program
database
info:eu-repo/classification/udc/004.4:004.896:004.925.84(043.2)
sami
Online Access:https://repozitorij.uni-lj.si/IzpisGradiva.php?id=148467
https://repozitorij.uni-lj.si/Dokument.php?id=172715&dn=
https://repozitorij.uni-lj.si/Dokument.php?id=172716&dn=
https://plus.cobiss.net/cobiss/si/sl/bib/163010051
https://hdl.handle.net/20.500.12556/RUL-148467
Description
Summary:Industrija tem bolj stremi k temu, da roboti opravljajo vse naloge. Glaven problem pa je, da se roboti ne znajo programirat sami in prepoznavati predmetov in okolice brez kakršnega koli treniranja. Tu nastopi ta zaključna naloga, katere namen je razvoj programske opreme za ustvarjanje slik iz CAD modelov. S pomočjo velike baze slik se lahko robot nauči prepoznavanja predmetov v okolici. Problema smo se lotili s programiranjem programa v programskem paketu Blender. Program sam generira slike določene resolucije, barvne globine in nastavljiv je potek in količina slik. Za vsako sceno program naredi dve sliki, ena je realistična in druga s črnim ozadjem. The industry is increasingly aiming for robots to perform all tasks. The main problem is that robots cannot program themselves and recognize objects and surroundings without any training. This is where this final assignment comes in, the purpose of which is to develop software for creating images from CAD models. With the help of a large database of images, the robot can learn to recognize objects in its surroundings. We tackled the problem by programming the program in the Blender software package. The program itself generates images of a certain resolution, color depth, and the sequence and quantity of images can be adjusted. For each scene, the program creates two images, one realistic and the other with a black background.

Similar Items