HMR-Net: Hierarchical Modular Routing for Cross-Domain Object Detection in Aerial Images

TL;DR

HMR-Net introduces a hierarchical modular routing framework for aerial object detection, enabling dataset-specific specialization and open-category detection without retraining, improving cross-domain generalization.

Contribution

The paper presents a novel hierarchical routing mechanism with dataset and scene-specific modules, and a conditional expert for open-category detection, advancing modular learning in aerial imagery analysis.

Findings

Improved multi-dataset generalization performance.

Enhanced regional specialization within scenes.

Enabled detection of novel object categories without retraining.

Abstract

Despite advances in object detection, aerial imagery remains a challenging domain, as models often fail to generalize across variations in spatial resolution, scene composition, and semantic label coverage. Differences in geographic context, sensor characteristics, and object distributions across datasets limit the capacity of conventional models to learn consistent and transferable representations. Shared methods trained on such data tend to impose a unified representation across fundamentally different domains, resulting in poor performance on region-specific content and less flexibility when dealing with novel object categories. To address this, we propose a novel modular learning framework that enables structured specialization in aerial detection. Our method introduces a hierarchical routing mechanism with two levels of modularity: a global expert assignment layer that uses latent geographic embeddings to route datasets to specialized processing modules, and a local scene decomposition mechanism that allocates image subregions to region-specific sub-modules. This allows our method to specialize across datasets and within complex scenes. Additionally, the framework contains a conditional expert module that uses external semantic information (e.g., category names or textual descriptions) to enable detection of novel object categories during inference, without the need for retraining or fine-tuning. By moving beyond monolithic representations, our method offers an adaptive framework for remote sensing object detection. Comprehensive evaluations on four datasets highlight improvements in multi-dataset generalization, regional specialization, and open-category detection.