Assessing the transferability of road network extraction methods to Slovenian orthophotos for power distribution network planning

Križaj, Janez (author), Knez, Klemen (author), Štruc, Vitomir (author)

In spatial planning of power distribution networks, road corridors provide an important basis for routing medium- and low-voltage power cables, as they often simplify permitting procedures, construction, and maintenance access. Automatic road extraction from aerial and satellite imagery is therefore not only a segmentation task, but also an input to subsequent optimization procedures, where connected and navigable road networks are required. The paper compares six recent methods for road network extraction: SAM-Road, D-LinkNet, CRESI, CU-dGCN, Sat2Graph, and U-Net-ResNet18. They are first evaluated on the SpaceNet dataset and then tested for transfer to Slovenian GURS orthophoto imagery. The results show that methods with an explicit graph output generally preserve road network connectivity better in the experiments. When transferred to Slovenian GURS orthophotos, their relative performance is largely retained. The methods that perform well on the SpaceNet also tend to perform well in this setting. A decline in their performance is observed mainly in rural areas, where roads are often narrow, partly occluded by vegetation or shadows, and visually less distinguishable from the surrounding terrain. Besides a comparative evaluation, the paper therefore provides practical guidance for selecting methods when constructing road networks for the optimization-supported power distribution network planning.

• Language: English
• Keywords: road extraction, benchmarking, satellite imagery, SpaceNet, domain shift, power distribution, spatial planning
• Work type: Article
• Typology: 1.01 - Original Scientific Article
• Organization: FE - Faculty of Electrical Engineering
• Publication version: Version of Record
• Publication date: 01.01.2026
• Year: 2026
• Number of pages: Str. 179-186
• Numbering: Letn. 93, št. 4
• PID: 20.500.12556/RUL-189978
• UDC: 621.31:004.93
• ISSN on article: 0013-5852
• COBISS.SI-ID: 300914691
• Publication date in RUL: 19.10.2026

Record is a part of a journal

• Title: Elektrotehniški vestnik
• Publisher: Strokovna zadruga koncesijoniranih elektrotehnikov, Elektrotehniška zveza Slovenije
• ISSN: 0013-5852
• COBISS.SI-ID: 742916

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.

Secondary language

• Language: Slovenian
• Title: Vrednotenje prenosljivosti metod za ekstrakcijo cestnih omrežij na slovenske ortofoto posnetke za potrebe načrtovanja distribucijskih omrežij

Abstract

Pri prostorskem načrtovanju elektrodistribucijskih omrežij so cestni koridorji pomembna podlaga za izbiro tras srednje- in nizkonapetostnih vodov, saj pogosto omogočajo lažje pridobivanje dovoljenj, enostavnejšo gradnjo in boljši dostop za vzdrževanje. Zato avtomatska ekstrakcija cest iz letalskih in satelitskih posnetkov ni uporabna le kot segmentacijska naloga, temveč predvsem kot vhod v nadaljnje postopke optimizacije, kjer so potrebna povezana in usmerljiva cestna omrežja. V članku primerjamo šest sodobnih pristopov za ekstrakcijo cestnih omrežij: SAM-Road, D-LinkNet, CRESI, CU-dGCN, Sat2Graph in U-Net-ResNet18. Metode najprej ovrednotimo na zbirki SpaceNet, nato pa preverimo še njihov prenos na slovenske ortofoto posnetke GURS-a. Rezultati kažejo, da metode z eksplicitnim grafnim izhodom v eksperimentih praviloma bolje ohranjajo povezanost cestnega omrežja. Pri prenosu na slovenske ortofoto posnetke GURS-a se relativna uspešnost metod večinoma ohrani, saj se pristopi, ki dosegajo dobre rezultate na zbirki SpaceNet, praviloma dobro izkažejo tudi v tem okolju. Poslabšanje je opazno predvsem na podeželskih območjih, kjer so ceste pogosto ožje, delno zakrite z vegetacijo ali sencami ter vizualno manj ločljive od okoliškega terena. Prispevek zato poleg primerjalnega vrednotenja podaja tudi praktične smernice za izbiro metod pri gradnji cestnega omrežja, uporabnega za optimizacijsko podprto načrtovanje elektrodistribucijskega omrežja.

• Keywords: ekstrakcija cest, primerjalno vrednotenje, satelitski posnetki, SpaceNet, premik domene, prostorsko načrtovanje, distribucijskega omrežja

Projects

• Funder: ARIS - Slovenian Research and Innovation Agency
• Project number: J2-60030
• Name: Optimalno umeščanje elektroenergetske infrastrukture v prostor s pomočjo umetne inteligence (OPTI-AI)