With the publication of DINO, a variant of the Detection Transformer (DETR), Detection Transformers are breaking the record in the object detection benchmark with the merits of their end-to-end design and scalability. However, the extension of DETR to oriented object detection has not been thoroughly studied although more benefits from its end-to-end architecture are expected such as removing NMS and anchor-related costs. In this paper, we propose a first strong DINO-based baseline for oriented object detection. We found that straightforward employment of DETRs for oriented object detection does not guarantee non-duplicate prediction, and propose a simple cost to mitigate this. Furthermore, we introduce a $\textit{dynamic denoising}$ strategy that uses Hungarian matching to filter redundant noised queries and $\textit{query alignment}$ to preserve matching consistency between Transformer decoder layers. Our proposed model outperforms previous rotated DETRs and other counterparts, achieving state-of-the-art performance in DOTA-v1.0/v1.5/v2.0, and DIOR-R benchmarks.
翻译:随着检测变换器变体DINO的发布,检测变换器凭借其端到端设计与可扩展性优势持续刷新目标检测基准记录。然而,尽管DETR的端到端架构有望在定向目标检测中带来更多益处(如消除非极大值抑制与锚点相关计算成本),但其在该领域的扩展尚未得到充分研究。本文首次提出基于DINO的强基线模型用于定向目标检测。研究发现,直接采用DETR系列模型进行定向目标检测无法保证无重复预测,为此我们提出一种简单代价函数以缓解该问题。进一步,我们引入$\textit{动态去噪}$策略,通过匈牙利匹配滤除冗余噪声查询,并采用$\textit{查询对齐}$机制保持变换器解码器层间的匹配一致性。所提模型性能超越现有旋转DETR方法及其他同类模型,在DOTA-v1.0/v1.5/v2.0与DIOR-R基准测试中均达到最优表现。