RBMatch: Dual-Level Class Rebalancing for Semi-Supervised Building Footprint Extraction

Akil Ahmad Taki; Shaikh Anowarul Fattah

cs.CVOct 6, 2026

RBMatch: Dual-Level Class Rebalancing for Semi-Supervised Building Footprint Extraction

Authors: Akil Ahmad Taki, Shaikh Anowarul Fattah

Organizations: Department of EEE, Bangladesh University of Engineering and Technology(BUET) · Department of EEE, University of Asia Pacific

Abstract

Accurate building footprint extraction from high-resolution remote sensing imagery is essential for urban planning, disaster response, and environmental monitoring. However, obtaining dense pixel-level annotations is costly, motivating the use of semi-supervised learning (SSL) to leverage unlabeled imagery. In remote sensing, severe foreground--background imbalance poses a particular challenge for self-training, as it can bias pseudo-label generation and the resulting unsupervised optimization toward the majority background class. We show that addressing this imbalance at only one stage is insufficient: balancing pseudo-label selection alone does not prevent background bias from re-emerging during unsupervised loss optimization, a failure mode we term \emph{imbalance leak}. To address this issue, we propose \textbf{RBMatch}, a dual-level class-rebalancing framework that jointly regulates pseudo-label generation and unsupervised optimization. RBMatch combines a supervised learning pathway with a self-training module comprising three components: adaptive class-specific thresholding (ACT) for balanced pseudo-label selection, confidence-aware class-balanced reweighting (CACBR) for mitigating class bias in the unsupervised loss, and distribution alignment (DAL) for matching the predicted unlabeled-data distribution to the labeled-data prior. Experiments on the WHU, INRIA, and Massachusetts building footprint datasets across labeled ratios of 1%--10% show that RBMatch consistently achieves the best building IoU and F1-score among the evaluated methods. The improvement is most pronounced on the highly imbalanced Massachusetts dataset, where RBMatch improves IoU by 1.37 points over the strongest baseline at a 1% labeling ratio and is the only method to outperform the fully supervised baseline across all twelve dataset--ratio settings.

Figures & tables

Fig. 1: Proposed RBMatch architecture. A shared SegFormer-B2 backbone processes the labeled stream and the weak/strong unlabeled views. Weak-view pseudo-labels (gradient stopped) drive ACT (adaptive thresholding), DAL (distribution alignment, anchored by $\pi^{l}$ ), and CACBR (confidence-aware reweighting), which produce per-pixel weights $w_{ij}$ for the reweighted unsupervised loss on the strong view.

Fig. 2: Empirical demonstration of the imbalance leak on WHU ( $r\!=\!0.01$ ). RBMatch’s gradient-weighted effective ratio corrects FreeMatch/FixMatch bias.

Fig. 3: Qualitative comparison on WHU ( $r\!=\!0.01$ ). White: TP, red: FP, blue: FN, black: TN. RBMatch recovers complete interiors without inflating false positives.

	Configuration	Recall	Precision	IoU	F1
(a)	Only-Sup	89.68	93.43	84.36	91.52
(b)	FixMatch (SUP+ST)	88.55	93.14	83.13	90.79
(c)	+ ACT only	88.60	94.04	83.89	91.24
(d)	+ CACBR only	88.58	94.04	83.87	91.23
(e)	ACT+CACBR (no DAL)	89.31	93.80	84.33	91.50
(f)	RBMatch (full)	90.21	93.88	85.19	92.00

TABLE I: Ablation on WHU at

r\!=\!0.05

. Best in bold .

WHU Building Dataset
Method	$r$ =0.01	$r$ =0.02	$r$ =0.05	$r$ =0.10
Only-Sup	83.27/90.87	83.55/91.04	84.36 / 91.52	85.38 / 92.11
FixMatch	82.42/90.36	82.89/90.64	83.13/90.79	84.74/91.74
FreeMatch	81.79/89.98	82.05/90.14	82.80/90.59	83.94/91.27
AdaptMatch	82.62/90.48	83.21/90.84	83.31/90.90	84.88/91.82
DARS	82.55/90.44	83.41/90.95	83.16/90.81	84.43/91.56

TABLE II: IoU/F1 (%) across datasets. Best in bold ; second-best underlined .

Method	UNet	UNet++	DeepLabV3	SegFormer-B2
Only-Sup	73.16/84.50	75.37/85.95	78.18/87.75	84.36/91.52
DBMatch	75.38/85.96	77.83/87.53	78.51/87.96	83.90/91.25
RBMatch	77.19/87.13	79.37/88.50	80.20/89.01	85.19/92.00

TABLE III: Generality across backbones on WHU at

r\!=\!0.05

. Best in bold.

Explore similar work

Sep 29, 2026cs.CV

UniBuild: Unified Building Mapping From Multi-Source Optical Remote Sensing Imagery With Detail Decoding and Geometry Regularization

Building extraction from optical remote sensing (RS) imagery is fundamental to urban mapping, yet existing methods are often dataset-specific and generalize poorly to unseen domains. Their practical use is also limited by insufficient detail recovery and weak geometric regularization, leading to blurred boundaries, irregular shapes, and merged adjacent buildings. To address these issues, we propose UniBuild, a unified building extraction framework for multi-source RGB optical RS imagery. First, a unified multi-dataset training scheme is constructed over heterogeneous RGB optical datasets to learn transferable building representations across sensors and resolutions. Second, a novel detail-preserving HR-DPT decoder is designed to integrate high-level semantic features with high-resolution spatial features, enhancing building detail recovery. Third, geometry-aware regularization is introduced through a structure-tensor-based direction-aware loss for boundary direction consistency and a saddle-aware loss for suppressing false activations in narrow inter-building gaps under low-resolution conditions. We train and evaluate UniBuild on multi-source RGB optical datasets, including 10 public high-resolution datasets and two self-collected low-resolution datasets. Experiments show that UniBuild consistently improves building-region accuracy, boundary sharpness, and adjacent-building separation across diverse datasets. It also generalizes well to unseen domains and supports practical building extraction from RGB optical RS imagery up to 10,m resolution. The predicted masks can be further converted into GIS-compatible building footprints through simple polygonization. The trained model and inference code are released at https://github.com/zhu-xlab/UniBuild.

Wei Huang, Chenying Liu, Yilei Shi +1

Chair of Data Science in Earth Observation, Technical University of Munich, 80333 Munich, Germany

Building Information Modeling Remote Sensing Geometry-Aware Regularization

Ask PDF

Apr 16, 2026cs.CV

Building Extraction from Remote Sensing Imagery under Hazy and Low-light Conditions: Benchmark and Baseline

Building extraction from optical Remote Sensing (RS) imagery suffers from performance degradation under real-world hazy and low-light conditions. However, existing optical methods and benchmarks focus primarily on ideal clear-weather conditions. While SAR offers all-weather sensing, its side-looking geometry causes geometric distortions. To address these challenges, we introduce HaLoBuilding, the first optical benchmark specifically designed for building extraction under hazy and low-light conditions. By leveraging a same-scene multitemporal pairing strategy, we ensure pixel-level label alignment and high fidelity even under extreme degradation. Building upon this benchmark, we propose HaLoBuild-Net, a novel end-to-end framework for building extraction in adverse RS scenarios. At its core, we develop a Spatial-Frequency Focus Module (SFFM) to effectively mitigate meteorological interference on building features by coupling large receptive field attention with frequency-aware channel reweighting guided by stable low-frequency anchors. Additionally, a Global Multi-scale Guidance Module (GMGM) provides global semantic constraints to anchor building topologies, while a Mutual-Guided Fusion Module (MGFM) implements bidirectional semantic-spatial calibration to suppress shallow noise and sharpen weather-induced blurred boundaries. Extensive experiments demonstrate that HaLoBuild-Net significantly outperforms state-of-the-art methods and conventional cascaded restoration-segmentation paradigms on the HaLoBuilding dataset, while maintaining robust generalization on WHU, INRIA, and LoveDA datasets. The source code and datasets are publicly available at: https://github.com/AeroVILab-AHU/HaLoBuilding.

Feifei Sang, Wei Lu, Hongruixuan Chen +2

MOE Key Lab of ICSP, Anhui Provincial Key Lab of Multimodal Cognitive Computation, IMIS Lab of Anhui Province, School of Computer Science and Technology, Anhui University, Hefei, China · Graduate School of Frontier Sciences, The University of Tokyo, Chiba, 277-8561, Japan

Single Image Dehazing Remote Sensing Low-Light Images Extraction Topology

Ask PDF

Jun 9, 2026cs.CV

Spatially Selective Self-Training for Unsupervised Building Change Detection

Unsupervised building change detection aims to learn building-change masks from unlabeled bi-temporal remote sensing images. Existing label-free methods often follow a discrepancy-to-mask paradigm, directly using temporal differences, frozen foundation-model responses, prompt-based outputs, or post-processing results as final change maps. Although these strategies provide annotation-free cues, they do not learn a task-specific building-change detector and remain vulnerable to the gap between generic temporal discrepancies and building-defined structural changes. In practice, such discrepancies are often noisy and task-irrelevant, as appearance shifts, registration errors, and non-building modifications can produce strong but misleading responses. To address this problem, we propose SST-CD, a spatially selective self-training framework that reformulates fully label-free building change detection as end-to-end detector learning under noisy pseudo supervision. SST-CD uses temporal discrepancies as candidate pseudo labels and trains the detector only on spatially reliable pixels, whose reliability is estimated by a local consistency criterion that filters inconsistent regions from supervision. To further stabilize noisy self-training, a lightweight feature adapter recalibrates bi-temporal features, while a prototype-based decoder produces compact change and no-change representations. Experiments on LEVIR-CD, WHU-CD, and DSIFN-CD show that SST-CD achieves F1 scores of 83.08%, 91.69%, and 86.60%, respectively, outperforming existing unsupervised and label-free baselines.

Wafaa I. M. Hussin, Zhi Lu, Anas M. I. Mohammed +3

School of Information and Communication Engineering, University of Electronic Science and Technology of China, Chengdu 611731, China · Laboratory of Intelligent Collaborative Computing, University of Electronic Science and Technology of China, Chengdu 611731, China · School of Civil Engineering, University of Khartoum, Khartoum, Sudan +2

Change Detection Unsupervised Detection Discrepancy

Ask PDF