HISTonAIR: A Benchmark Dataset for Land-Cover Segmentation From Historical Aerial Imagery in Norway
IEEE Geoscience and Remote Sensing Letters · 2026 · 23 · 1-5
DOI: 10.1109/lgrs.2026.3678859 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Geoscience and Remote Sensing Letters, 2026. Verfasst von Mahmoud Saber Kenawi und 4 weiteren Autor:innen. Die Arbeit nennt 20 Literaturangaben. Gefördert von: Project Footprint and Impacts of Renewable Energy: Pressure on Lands Under Growth, Research Council of Norway.
Abstract (Originalsprache)
This work presents a very high-resolution (VHR) land-cover dataset derived from historical grayscale aerial imagery over Norwegian hydropower catchments. Addressing the lack of publicly documented, large-scale labeled data for legacy aerial images, the dataset comprises approximately 3.46 million spatially aligned image–label patches annotated across six land-cover classes. A benchmark subset of 15% from the dataset was used to validate segmentation performance using convolutional neural network (CNN) architectures, including U-Net and DeepLabV3, with ResNet-101 and EfficientNet-B7 encoders. Despite limited spectral information, the dataset supported consistent and competitive results with $F1$ -score up to 0.77, confirming its suitability for scalable deep learning applications. The work enables reproducible benchmarking on monochromatic imagery and supports future research in environmental reconstruction, land-use change analysis, and domain adaptation. The dataset, preprocessing tools, and evaluation scripts are made publicly available to support community use and model development.
Förderung
- Project Footprint and Impacts of Renewable Energy: Pressure on Lands Under Growth
- Research Council of Norway
Zitieren
APA
Kenawi, M. S., Vakhidi, F., Alfredsen, K., Bakken, T. H., & Ullo, S. L. (2026). HISTonAIR: A Benchmark Dataset for Land-Cover Segmentation From Historical Aerial Imagery in Norway. IEEE Geoscience and Remote Sensing Letters, 23, 1-5. https://doi.org/10.1109/lgrs.2026.3678859
Vancouver
Kenawi MS, Vakhidi F, Alfredsen K, Bakken TH, Ullo SL. HISTonAIR: A Benchmark Dataset for Land-Cover Segmentation From Historical Aerial Imagery in Norway. IEEE Geoscience and Remote Sensing Letters. 2026;23:1-5. doi:10.1109/lgrs.2026.3678859
BibTeX
@article{Kenawi2026,
author = {Kenawi, Mahmoud Saber and Vakhidi, Fabian and Alfredsen, Knut and Bakken, Tor H. and Ullo, Silvia Liberata},
title = {HISTonAIR: A Benchmark Dataset for Land-Cover Segmentation From Historical Aerial Imagery in Norway},
journal = {IEEE Geoscience and Remote Sensing Letters},
year = {2026},
volume = {23},
pages = {1--5},
doi = {10.1109/lgrs.2026.3678859}
}
EUVIMED