Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

STAR-IOD: Scale-decoupled Topology Alignment with Pseudo-label Refinement for Remote Sensing Incremental Object Detection

About

Remote sensing imagery typically arrives in the form of continuous data streams. Traditional detectors often forget previously learned categories when learning new ones; therefore, research on Remote Sensing Incremental Object Detection (RS-IOD) is of great significance. However, existing methods largely overlook the intra-class scale variations prevalent in remote sensing scenes, which undermines the effectiveness of knowledge transfer and old knowledge preservation. Moreover, RS-IOD also suffers from missing annotations, which cause the model to misclassify old-class instances as background. To address these challenges, we propose a novel framework, STAR-IOD. First, we introduce a Subspace-decoupled Topology Distillation (STD) module to transfer structural knowledge, explicitly aligning inter-class topological relationships and mitigating intra-class representation discrepancies induced by scale shifts. Furthermore, we introduce the Clustering-driven Pseudo-label Generator (CPG), a plug-and-play module that leverages K-Means clustering to dynamically identify class-specific thresholds, thereby guaranteeing an accurate distinction between true positive targets and background noise and alleviating the issue of missing annotations for old classes. We also constructed two Remote Sensing Incremental Object Detection datasets, DIOR-IOD and DOTA-IOD to facilitate research on RS-IOD. Extensive experiments demonstrate that our method outperforms state-of-the-art approaches by 1.7% and 2.1% mAP on DIOR-IOD and DOTA-IOD, respectively, effectively alleviating catastrophic forgetting while preserving strong detection performance on both base and novel classes. The code and dataset are released at: https://github.com/zyt95579/STAR-IOD.

Yaoteng Zhang, Qing Zhou, Junyu Gao, Qi Wang• 2026

Related benchmarks

TaskDatasetResultRank
Incremental Object DetectionDOTA-IOD Task 2
mAP (All)42.4
6
Incremental Object DetectionDOTA-IOD Task 3
mAP^A39.4
6
Incremental Object DetectionDOTA Task 1
mAP^C50.9
6
Incremental Object DetectionDIOR-IOD two-step
mAP^C45.6
6
Object DetectionDIOR-IOD multi-step Task 2
mAP^A43
5
Object DetectionDIOR-IOD multi-step Task 3
mAP^A37.6
5
Object DetectionDIOR-IOD multi-step Task 4
mAP^A36.9
5
Object DetectionDIOR-IOD multi-step Task 1
mAP^A44
5
Showing 8 of 8 rows

Other info

Follow for update