MARSS：基于模块化注意力与状态空间模型的雷达语义分割
MARSS: Radar Semantic Segmentation via Modular Attention and State Space Models
面向雷达距离-多普勒频率图这类各向异性、稀疏含噪输入，给出可复用的模块化设计：用轻量通道自注意力+深度可分离卷积编码噪声特征、用区域级注意力做多尺度融合、用状态空间模型结合轴向自注意力在解码端建模各向异性与时序一致性，可迁移到其他非相机模态的密集预测任务。
原文依据 the decoder's RADM module combines state space models with axial self-attention to reconstruct segmentation masks
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论雷达语义分割（RSS）对于恶劣条件下的鲁棒感知至关重要，但带来了独特的挑战：雷达频率图具有高度各向异性、多尺度、稀疏且含噪的特点。为相机图像设计的传统 CNN 或 Transformer 架构未能考虑这些特性，导致性能下降。我们提出 MARSS（模块化注意力增强的雷达语义分割），这是一个集成三个专用模块以解决雷达特有问题的新框架。在编码器中，RADE 模块采用轻量通道自注意力与深度卷积，以鲁棒地编码含噪、各向异性的特征。在中间层，RFAF 模块执行多尺度特征融合与区域级注意力，以分离出显著的雷达特征。解码器的 RADM 模块将状态空间模型与轴向自注意力相结合，以各向异性与时序感知的上下文重建分割掩码。这些组件共同抑制噪声、解耦距离-多普勒特征，并强化时空一致性。在 CARRADA 数据集上，MARSS 取得了显著高于先前 RSS 方法的性能，尤其是对于小型快速移动目标。
英文原文摘要 1223 字符
Radar semantic segmentation (RSS) is critical for robust perception in adverse conditions, but poses unique challenges: radar frequency maps are highly anisotropic, multi-scale, sparse and noisy. Conventional CNN or Transformer architectures, designed for camera images, fail to account for these characteristics, leading degraded performance. We propose MARSS (Modular Attention-enhanced Radar Semantic Segmentation), a novel framework that integrates three specialized modules to address radar-specific issues. In the encoder, the RADE module employs lightweight channel self-attention and depthwise convolutions to robustly encode noisy, anisotropic features. In intermediate layers, the RFAF module performs multi-scale feature fusion and region-level attention to isolate salient radar features. The decoder's RADM module combines state space models with axial self-attention to reconstruct segmentation masks with anisotropy and temporality-aware context. These components collectively suppress noise, disentangle range-Doppler features, and enforce spatial-temporal consistency. On the CARRADA dataset, MARSS achieves substantially higher performance than prior RSS methods, especially for small fast-moving targets.