Insights into the Effects of Tile Size and Tile Overlap Levels on Semantic Segmentation Models Train...
Insights into the Effects of Tile Size and Tile Overlap Levels on Semantic Segmentation Models Trained for Road Surface Area Extraction from Aerial Orthophotography
About this item
Full title
Author / Creator
Publisher
Basel: MDPI AG
Journal title
Language
English
Formats
Publication information
Publisher
Basel: MDPI AG
Subjects
More information
Scope and Contents
Contents
Studies addressing the supervised extraction of geospatial elements from aerial imagery with semantic segmentation operations (including road surface areas) commonly feature tile sizes varying from 256 × 256 pixels to 1024 × 1024 pixels with no overlap. Relevant geo-computing works in the field often comment on prediction errors that could be attri...
Alternative Titles
Full title
Insights into the Effects of Tile Size and Tile Overlap Levels on Semantic Segmentation Models Trained for Road Surface Area Extraction from Aerial Orthophotography
Authors, Artists and Contributors
Identifiers
Primary Identifiers
Record Identifier
TN_cdi_doaj_primary_oai_doaj_org_article_420c56ed12274b4bb0484e9ce0fd9483
Permalink
https://devfeature-collection.sl.nsw.gov.au/record/TN_cdi_doaj_primary_oai_doaj_org_article_420c56ed12274b4bb0484e9ce0fd9483
Other Identifiers
ISSN
2072-4292
E-ISSN
2072-4292
DOI
10.3390/rs16162954
