LEVIRDet: A Million-Scale 159-Category Dataset and Foundation Model for Universal Remote Sensing Object Detection

2026-06-23

AI Summary

Existing remote sensing object detection datasets suffer from limited categories, fixed resolution, and reliance on a single sensor type, which hinders the cross-domain generalization of detection models. To address these limitations, this work introduces LEVIRDet-159, a large-scale, multi-granularity remote sensing detection benchmark comprising 159 object classes and 2.56 million annotated bounding boxes, along with a unified detection architecture named LEVIRDetNet. The proposed model integrates online visual ground sampling distance (GSD) prediction, GSD-conditioned query modulation, and a hierarchical-aware detection head. Remarkably, without any training or fine-tuning on target domains, LEVIRDetNet outperforms the strongest fully supervised methods by an average of 5.02 mAP across nine external benchmarks, significantly advancing generalization performance across diverse sensors and taxonomic systems.

Abstract

Remote sensing object detection has advanced rapidly with the development of large-scale benchmarks and modern detection architectures. However, existing datasets and detectors remain fragmented. Most benchmarks focus on limited categories, fixed spatial resolutions, or a single sensor, while detectors still struggle to work across different sensors and categorical systems. In this paper, we introduce LEVIRDet-159, the largest and most comprehensive remote sensing object detection dataset to date, with 159 categories, 2.56 million bounding boxes, and 700k fine-grained annotations under a multi-level taxonomy. In each key scale dimension, LEVIRDet-159 exceeds the corresponding largest existing remote sensing object detection dataset, containing approximately (7x) more images, (6x) more object instances, and (4x) more categories. Based on this dataset, we design LEVIRDetNet, a scale-hierarchy-aware detection foundation model for universal remote sensing object detection. LEVIRDetNet couples online visual Ground Sampling Distance (GSD) prediction, GSD-conditioned query modulation and allocation, and a hierarchy-aware detection head for mixed-granularity remote sensing supervision. Under stringent evaluation settings, LEVIRDetNet demonstrates strong cross-domain generalization. Even without target-domain training or fine-tuning, it achieves state-of-the-art detection performance on 9 external benchmarks, improving the strongest fully supervised competing methods by 5.02 mAP on average under each benchmark's primary metric. We hope this study will facilitate the development of strongly generalizable remote sensing object detection across diverse category systems, spatial resolutions, and sensor platforms. The dataset and trained models will be released at https://qinzheyang.github.io/LEVIRDet/, accompanying the final paper.

Problem

• remote sensing object detection
• dataset fragmentation
• cross-sensor generalization
• category scalability
• spatial resolution variability

Innovation

• remote sensing object detection
• foundation model
• scale-hierarchy-aware
• cross-domain generalization
• multi-level taxonomy