End-to-End Monocular Building Height Estimation with Semantic Interval Guidance and Region-Aware Modulation
Authors:
- Ruijie HanRHRuijie Han
- Xiangtao FanXFXiangtao Fan
- Jian LiuJian Liu
- Wenjuan ZhangWZWenjuan Zhang
- Weijia BeiWBWeijia Bei
- Qifeng GeQGQifeng Ge
- Jianhao XuJXJianhao Xu
- Ruijie YaoRYRuijie Yao
Preprint
- 1 June 2026
- preprint
- Published by Elsevier in SSRN Electronic Journal
Abstract
Accurate height estimation from single high-resolution satellite imagery remains a fundamental yet challenging problem in urban remote sensing. The primary difficulties stem from three aspects: insufficient cross-scale perception leads to blurred building boundaries; limited association between height interval modeling and global height semantics results in incomplete recovery of rooftop structural details; and uniform spatial modulation causes predictions in extreme-height regions to converge toward the mean, degrading local accuracy. To address these challenges, this paper proposes a novel monocular height estimation framework built upon a height-binned modeling paradigm and progressively optimized from global feature encoding to local spatial refinement. Specifically, a bidirectional multi-scale feature aggregation mechanism enhances cross-scale spatial perception; a semantic-guided height interval modeling module incorporates global height semantics into bin representations for more discriminative interval prediction; and a region-aware pixel-level modulation mechanism explicitly models height interval-spatial region relationships to suppress extreme-height prediction bias. Extensive experiments on DFC2023 and US3D benchmark datasets demonstrate consistently superior performance over all compared methods. Validation on GF-7 domestic satellite imagery further confirms the framework's generalization to distinct radiometric and imaging characteristics, demonstrating its practical value for large-scale domestic 3D urban mapping.Keywords
This publication has 12 references indexed in Scilit:
- 3D building reconstruction from monocular remote sensing imagery via diffusion models and geometric priorsISPRS Journal of Photogrammetry and Remote Sensing, 2025
- HeightFormer: Single-Imagery Height Estimation Transformer With Bilateral Feature Pyramid FusionIEEE Geoscience and Remote Sensing Letters, 2024
- Effects of building height on the sound transmission in cross-laminated timber buildings – Airborne sound insulationBuilding and Environment, 2023
- Rethinking Monocular Height Estimation From a Classification Task Perspective Leveraging the Vision TransformerIEEE Geoscience and Remote Sensing Letters, 2022
- Shadow Pattern-Enhanced Building Height Extraction Using Very-High-Resolution ImageIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2022
- Deep Neural Networks for Determining the Parameters of Buildings from Single-Shot Satellite ImageryJournal of Computer and Systems Sciences International, 2020
- U-IMG2DSM: Unpaired Simulation of Digital Surface Models With Generative Adversarial NetworksIEEE Geoscience and Remote Sensing Letters, 2020
- Towards depth estimation in a single aerial imageInternational Journal of Remote Sensing, 2019
- Generating 3D city models without elevation dataComputers, Environment and Urban Systems, 2017
- Global scenarios of urban density and its impacts on building energy use through 2050Proceedings of the National Academy of Sciences, 2017