What does this research mean for the field?

Integrating Kolmogorov-Arnold Network (KAN) layers into a U-Net architecture (KANU-Net) enables accurate monocular height estimation from single optical images in heterogeneous urban environments. Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

What question did this study set out to answer?

This research aims to improve height estimation from single optical images through a novel architecture, KANU-Net.

May 31, 2026ISPRS annals of the photogrammetry, remote sensing and spatial information sciencesOpen Access

Height Estimation from Single Optical Images Using KANU-Net Architecture

View Full Paper

Ask AI

Bookmark

Authors

RVReyhaneh Vahabi HAHossein Arefi RBReza Bahmanyar

Discussion

Member takes

Overview

Randomized trial demonstrates monocular height estimation accuracy in urban areas, suggesting a new approach for 3D reconstruction.

Key Points

This research aims to improve height estimation from single optical images through a novel architecture, KANU-Net.
Developed KANU-Net, integrating KAN layers for enriched feature representation.
Processed high-resolution optical images into 256x256 patches for evaluation.
Conducted qualitative and quantitative assessments in two urban areas: Utrecht and Potsdam.
Achieved RMSE values of 3.43 m for Utrecht and 3.29 m for Potsdam.
Reported accuracy rates (δ₁) over 0.43 for Utrecht and over 0.50 for Potsdam.
Demonstrated the model's ability to produce detailed and consistent height maps across urban morphologies.

Cite This Study

Vahabi et al. (2026) studied this question.

synapsesocial.com/papers/6a1bd1555783ba022b6fcf4f https://doi.org/10.5194/isprs-annals-x-4-w8-2025-809-2026

View Full Paper

Ask AI

Bookmark

May 31, 2026ISPRS annals of the photogrammetry, remote sensing and spatial information sciencesOpen Access

Height Estimation from Single Optical Images Using KANU-Net Architecture

View Full Paper

Ask AI

Bookmark

Authors

RVReyhaneh Vahabi HAHossein Arefi RBReza Bahmanyar

Discussion

Member takes

Overview

Randomized trial demonstrates monocular height estimation accuracy in urban areas, suggesting a new approach for 3D reconstruction.

Key Points

This research aims to improve height estimation from single optical images through a novel architecture, KANU-Net.
Developed KANU-Net, integrating KAN layers for enriched feature representation.
Processed high-resolution optical images into 256x256 patches for evaluation.
Conducted qualitative and quantitative assessments in two urban areas: Utrecht and Potsdam.
Achieved RMSE values of 3.43 m for Utrecht and 3.29 m for Potsdam.
Reported accuracy rates (δ₁) over 0.43 for Utrecht and over 0.50 for Potsdam.
Demonstrated the model's ability to produce detailed and consistent height maps across urban morphologies.

Cite This Study

Vahabi et al. (2026) studied this question.

synapsesocial.com/papers/6a1bd1555783ba022b6fcf4f https://doi.org/10.5194/isprs-annals-x-4-w8-2025-809-2026

View Full Paper

Ask AI

Bookmark