GRADE：视觉退化下的单帧生成式雷达深度估计
原标题：GRADE: Single-Frame Generative Radar Depth Estimation Under Visual Degradation
AI 摘要
该论文提出GRADE方法，利用预训练生成先验与单帧毫米波雷达几何信息，在烟雾、雾霾和黑暗等视觉退化条件下估计高保真度量深度。GRADE先将原始4D雷达频谱映射为粗略度量深度，再用潜在扩散模型恢复结构细节，并通过像素空间适配器在可见时利用相机残差线索。在12栋建筑95K帧数据上训练评估，清晰场景MAE为0.303米，烟雾下为0.313米，优于现有基线。
正文节选
GRADE: Single-Frame Generative Radar Depth Estimation Under Visual Degradation Abstract. Dense 3D depth perception fails under smoke, fog, and darkness because optical sensors cannot penetrate airborne particulates. mmWave radar remains usable and measures range accurately under these conditions, but its small aperture limits angular resolution. We present GRADE, which grounds a pretrained generative prior in single-frame radar geometry to estimate high-fidelity metric depth. GRADE first maps ra