Balanced Orthogonal Subspace Separation Detector for Few-Shot Object Detection in Aerial Imagery

Few-shot object detection (FSOD) in remote sensing images (RSIs) aims to achieve object location and classification with only few training samples. Currently, mainstream transfer-learning methods employ a two-stage approach: pretraining on data-abundant base classes and fine-tuning on few-shot novel classes. However, existing approaches suffer notable degradation in both base and novel classes during fine-tuning, because of gradients conflict and class imbalance. To address this, we construct the Balanced Orthogonal Subspace Separation Detector (BOSS), a novel two-stage framework for few-shot object detection. Specifically, to avoid contradictory gradients, BOSS distinctly isolates the training of base and novel classes at both structural and feature levels. For structural separation, a low-rank subspace adapter is introduced to ensure network optimization for novel classes without hampering base classes pretraining performance, effectively addressing over-fitting in few-shot scenarios. For feature disentanglement, an orthogonal subspace extractor is presented, enhancing class separability by learning class-specific, orthogonal basis-spanned subspace. Finally, a balanced classifier is proposed to equalize the imbalanced loss, with its dual-component design mitigating bias towards predicting background or base classes. Comparative evaluations on diverse remote sensing datasets demonstrate BOSS’s superiority, outperforming state-of-the-art methods in mAP. These results underscore BOSS’s effectiveness in few-shot object detection, particularly in challenging remote sensing contexts.