Uporaba konvolucijskih nevronskih mrež za detektiranje krošenj dreves

Škrlj, Janez

| | SLO | ENG | Cookies and privacy

Bigger font | Smaller font

First page > Show document

Show document

Title:	Uporaba konvolucijskih nevronskih mrež za detektiranje krošenj dreves
Authors:	ID Škrlj, Janez (Author) ID Potočnik, Božidar (Mentor) More about this mentor...
Files:	VS_Skrlj_Janez_2025_(3).pdf (3,69 MB) MD5: 7A8A3AC54A4FE9F2737863F37B464FD3
Language:	Slovenian
Work type:	Bachelor thesis/paper
Typology:	2.11 - Undergraduate Thesis
Organization:	FERI - Faculty of Electrical Engineering and Computer Science
Abstract:	Diplomsko delo se ukvarja z razvojem računskega modela za detektiranje krošenj dreves v tlorisni sliki gozda. Na osnovi študija literature smo ugotovili, da je problem najlažje rešljiv z uporabo konvolucijskih nevronskih mrež. Za praktični del tega zaključnega dela smo izbrali prepoznavalnik YOLO (angl. You Only Look Once). Svojo rešitev z uporabo različice YOLO v7 smo implementirali v orodju DarkNet. Pripravili smo lastno zbirko slik, ki smo jih zajeli s pomočjo brezpilotnega letalnika. Množico podatkov smo ustrezno razdelili v učno in testno množico. Analiza dobljenih rezultatov je pokazala, da je naša rešitev zmožna prepoznavati krošnje dreves v tlorisnih slikah gozdov. Metrika natančnost je bila nad 90 %, je pa bila metrika priklic nekoliko nižja, tj. okrog 50 %. Diplomsko delo podaja tudi nekaj smernic za uporabo tako naučenih modelov v gozdarstvu.
Keywords:	strojno učenje, konvolucijske nevronske mreže, model “You Only Look Once“ (YOLO), računalniški vid, prepoznava objektov s slike
Place of publishing:	Maribor
Publisher:	[J. Škrlj]
Year of publishing:	2025
PID:	20.500.12556/DKUM-93447
UDC:	004.85.032.26:004.932(043.2)
COBISS.SI-ID:	21678
Publication date in DKUM:	01.10.2025
Views:	187
Downloads:	43
Metadata:
Categories:	KTFMB - FERI
:	Copy citation

Average score:	(0 votes)
Your score:	Voting is allowed only for logged in users.
Share:

Hover the mouse pointer over a document title to show the abstract or click on the title to get all document metadata.

Licences

License:	CC BY 4.0, Creative Commons Attribution 4.0 International

Link:	http://creativecommons.org/licenses/by/4.0/
Description:	This is the standard Creative Commons license that gives others maximum freedom to do what they want with the work as long as they credit the author.
Licensing start date:	26.06.2025

Secondary language

Language:	English
Title:	Using convolutional neural networks for detecting tree canopies
Abstract:	The thesis deals with the development of a computational model for the detection of tree canopy on the top-down image of a forest landscape. Based on a literature study, we found that the problem is most easily solved using convolutional neural networks. For the practical part of this thesis, we have chosen the YOLO (You only look once) recogniser. We have implemented our solution using YOLO v7 in the DarkNet toolbox. We have created our own collection of images captured by a drone. We split the dataset into training and test sets accordingly. The analysis of the obtained results showed that our solution is able to recognize tree canopy in top-down images of forests. The accuracy metric was above 90%, but the recall metric was slightly lower, i.e. around 50%. The thesis also gives some guidelines for the application of such learned models in forestry.
Keywords:	machine learning, convolutional neural networks, You Only Look Once (YOLO), computer vision, object recognition from images

Comments

Leave comment

You must log in to leave a comment.

Comments (0)

0 - 0 / 0

There are no comments!

Back