DA-Mamba：学习域感知状态空间模型用于域自适应目标检测中的全局-局部对齐
DA-Mamba: Learning Domain-Aware State Space Model for Global-Local Alignment in Domain Adaptive Object Detection
用线性复杂度的状态空间模型（Mamba）替代 Transformer 的二次注意力，在骨干网络和检测头中分别插入图像级与实例级的全局建模模块，为跨域目标检测提供了一种可复用的全局-局部对齐范式，也可迁移到其他需要长程依赖的密集预测任务。
原文依据 we propose DA-Mamba, a hybrid CNN-State Space Models (SSMs) architecture that combines the efficiency of CNNs with the linear-time long-range modeling capability
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论域自适应目标检测（DAOD）旨在将检测器从有标注的源域迁移到无标注的目标域。现有的 DAOD 方法采用多粒度特征对齐来学习域不变表示。然而，其基于 CNN 的骨干网络与检测头的局部连通性将对齐限制在局部区域，无法提取全局域不变特征。尽管基于 Transformer 的 DAOD 方法通过注意力机制捕获全局依赖，但其二次计算开销阻碍了实际部署。为解决这一问题，我们提出 DA-Mamba，一种混合 CNN-状态空间模型（SSM）架构，将 CNN 的高效性与状态空间模型的线性时间长程建模能力相结合，以同时捕获全局与局部的域不变特征。具体而言，我们引入两个新模块：图像感知 SSM（IA-SSM）与目标感知 SSM（OA-SSM）。IA-SSM 被集成到骨干网络中，以增强全局域感知能力，从而实现图像级的全局与局部对齐。OA-SSM 被插入检测头中，以建模目标之间的空间与语义依赖关系，从而增强实例级对齐。大量实验表明，所提方法能够高效提升目标检测器的跨域性能。
英文原文摘要 1325 字符
Domain Adaptive Object Detection (DAOD) aims to transfer detectors from a labeled source domain to an unlabeled target domain.Existing DAOD methods employ multi-granularity feature alignment to learn domain-invariant representations.However, the local connectivity of their CNN-based backbone and detection head restricts alignment to local regions, failing to extract global domain-invariant features.Although transformer-based DAOD methods capture global dependencies via attention mechanisms, their quadratic computational cost hinders practical deployment. To solve this, we propose DA-Mamba, a hybrid CNN-State Space Models (SSMs) architecture that combines the efficiency of CNNs with the linear-time long-range modeling capability of State Space Models (SSMs) to capture both global and local domain-invariant features.Specifically, we introduce two novel modules: Image-Aware SSM (IA-SSM) and Object-Aware SSM (OA-SSM).IA-SSM is integrated into the backbone to enhance global domain awareness, enabling image-level global and local alignment.OA-SSM is inserted into the detection head to model spatial and semantic dependencies among objects, enhancing instance-level alignment.Comprehensive experiments demonstrate that the proposed method can efficiently improve the cross-domain performance of the object detector.