Powered by OpenAIRE graph
Found an issue? Give us feedback
image/svg+xml art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos Open Access logo, converted into svg, designed by PLoS. This version with transparent background. http://commons.wikimedia.org/wiki/File:Open_Access_logo_PLoS_white.svg art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos http://www.plos.org/ NTNU Openarrow_drop_down
image/svg+xml art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos Open Access logo, converted into svg, designed by PLoS. This version with transparent background. http://commons.wikimedia.org/wiki/File:Open_Access_logo_PLoS_white.svg art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos http://www.plos.org/
NTNU Open
Bachelor thesis . 2024
Data sources: NTNU Open
addClaim

Semantic segmentation dashboard

Authors: Edvardsen, Petter; Ertshus, Sivert; Finstad, Jakob;

Semantic segmentation dashboard

Abstract

Arkwiz er et selskap som tilbyr realistiske 3D-modeller av miljøer ved hjelp av geospatiale data. Dataene hentes ut gjennom klassifisering av bakkedekke, en grunnleggende oppgave innen geografi og miljøvitenskap i mange år, som tradisjonelt har vært avhengig av manuell tolkning og enkle datateknikker. Imidlertid har den nylige trenden innen dette feltet skiftet mot å utnytte maskinlæring, spesielt avanserte klassifiseringsmodeller, for å forbedre nøyaktigheten og effektiviteten i denne prosessen. Oppdraget gitt av Arkwiz hadde som mål å utvikle en robust maskinlæringsmodell som er i stand til å klassifisere intrikate detaljer fra flyfoto for å lette skapelsen av virkelighetsnære 3D-miljøer. Denne oppgaven krevde at vi nøyaktig segmenterte ulike typer bakkedekke og strukturer innen bildene, og ga kritiske data for å forbedre realismen og nøyaktigheten av deres 3D-modeller. Vår tilnærming innebar å utnytte avanserte nevrale nettverksarkitekturer for å fullføre oppgaven. Gjennom hele prosjektet oppnådde vi betydelige forbedringer gjennom iterative forfininger og grundig forskning. Den resulterende segmenteringsmodellen gir verdifull innsikt ved nøyaktig å identifisere og klassifisere funksjoner fra luftbilder. Denne integrasjonen strømlinjeformer ikke bare arbeidsflyten, men gir også et kraftig verktøy for å utvikle detaljerte og nøyaktige 3D-miljøer. Etter hvert som prosjektet utviklet seg, utvidet vi omfanget til å inkludere utviklingen av et omfattende segmenteringsdashbord. Dette dashbordet fungerer som et grensesnitt for alle brukere til å samhandle med og visualisere modellens output, og tilbyr en intuitiv og tilgjengelig måte å analysere segmenteringsresultatene på. Det gir en rask og fleksibel løsning for å identifisere miljødata fra tidligere undermappede interesseområder, og gir betydelig verdi til Arkwiz.

Arkwiz is a company which offers realistic 3D models of environments using geospatial data. The data is extracted through land cover classification which has been a fundamental task in geographic and environmental sciences for many years, traditionally relying on manual interpretation and basic computational techniques. However, the recent trend in this field has shifted towards leveraging machine learning, specifically advanced classification models, to enhance accuracy and efficiency of this process. Tasked by Arkwiz, the objective of our project was to develop a robust machine learning model capable of classifying intricate details from aerial photos to facilitate the creation of real-world 3D environments. This task required us to accurately segment various land cover types and structures within the imagery, providing critical data for enhancing the realism and accuracy of their 3D models. Our approach involved leveraging advanced neural network architectures to accomplish the task. Throughout this project we achieved significant improvements through iterative refinements and thorough research. The resulting segmentation model delivers valuable insights by accurately identifying and classifying features from aerial images. This integration not only streamlines the workflow but also provides a powerful tool for developing detailed and accurate 3D environments. As the project progressed, we expanded our scope to include the development of a comprehensive segmentation dashboard. This dashboard serves as an interface for all users to interact with and visualize the model's output, offering an intuitive and accessible way to analyze the segmentation results. It provides a fast and flexible solution to identify environmental data from previously undermapped areas of interest providing significant value to Arkwiz.

Country
Norway
  • BIP!
    Impact byBIP!
    selected citations
    These citations are derived from selected sources.
    This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
    0
    popularity
    This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network.
    Average
    influence
    This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
    Average
    impulse
    This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network.
    Average
Powered by OpenAIRE graph
Found an issue? Give us feedback
selected citations
These citations are derived from selected sources.
This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
BIP!Citations provided by BIP!
popularity
This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network.
BIP!Popularity provided by BIP!
influence
This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
BIP!Influence provided by BIP!
impulse
This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network.
BIP!Impulse provided by BIP!
0
Average
Average
Average
Green