Search papers, labs, and topics across Lattice.
HMR-Net is introduced to address the challenge of cross-domain object detection in aerial images by enabling structured specialization. It uses a hierarchical routing mechanism with global expert assignment based on latent geographic embeddings and local scene decomposition to allocate image subregions to region-specific sub-modules. The framework also incorporates a conditional expert module that leverages external semantic information for detecting novel object categories without retraining.
Achieve state-of-the-art aerial object detection by learning specialized representations across diverse datasets and within complex scenes, and even detect novel object categories without retraining.
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.
