先修复再融合：面向损坏但仍存在的传感器的冻结宿主适配

中文摘要 AI 辅助

相机-激光雷达检测器即使在两种传感器均存在、同步且已标定的情况下，仍可能持续消费不可靠的特征。我们提出了边界特征修复（Boundary Feature Repair, BFR），这是一种冻结宿主适配框架，在检测器已消费的模态接口处学习任务监督的残差修正。BFR-C修复融合所读取的每个相机特征层级，而BFR-L将宿主条件下的激光雷达候选对齐到选定边界，并相对于冻结锚点路由站点级创新。它们的联合训练组合为BFR-CL。零初始化的逐通道缩放使得每个变体在优化前在检测器层面精确等同于恒等映射；仅修复模块进行训练，而编码器、融合消费者、路由器、检测头及宿主归一化统计量均保持固定。在推理时，BFR既不需要干净参考、损坏元数据、时间历史，也不需要在线更新。在完整的20种损坏、5种严重度的KITTI-C网格上，BFR-C将MVX-Net的RCE从14.07降至11.92，并将Focals Conv-F的RCE从14.29降至11.00，相对于其复现的冻结基线。在后一宿主机上，BFR-L将AP_cor从73.65提升至74.48，而BFR-CL达到77.01的AP_cor和10.46的RCE，同时保持86.02的干净AP。在nuScenes-R上，BFR-CL将复现的MoME基线的mAP鲁棒性比率从80.1提升至81.4。这些结果确立了边界修复作为一种针对损坏但仍存在的传感的有针对性改造，无需重新训练已部署的检测器。

英文摘要

Camera-LiDAR detectors can continue to consume unreliable features even when both sensors remain present, synchronized, and calibrated. We introduce \emph{Boundary Feature Repair} (BFR), a frozen-host adaptation framework that learns task-supervised residual corrections at modality interfaces the detector already consumes. BFR-C repairs each camera feature level read by fusion, whereas BFR-L aligns host-conditioned LiDAR candidates to a selected boundary and routes site-wise innovations relative to the frozen anchor. Their jointly trained composition is BFR-CL. Zero-initialized per-channel scales make every variant an exact detector-level identity before optimization; only the repair modules train, while the encoders, fusion consumer, router, detection head, and host normalization statistics remain fixed. At inference, BFR requires neither clean references, corruption metadata, temporal history, nor online updates. Across the complete 20-corruption, five-severity KITTI-C grid, BFR-C reduces RCE from $14.07$ to $11.92$ on MVX-Net and from $14.29$ to $11.00$ on Focals Conv-F relative to their reproduced frozen baselines. On the latter host, BFR-L raises AP$_{\mathrm{cor}}$ from $73.65$ to $74.48$, while BFR-CL reaches $77.01$ AP$_{\mathrm{cor}}$ and $10.46$ RCE with $86.02$ clean AP. On nuScenes-R, BFR-CL raises the reproduced MoME baseline's mAP robustness ratio from $80.1$ to $81.4$. These results establish boundary repair as a targeted retrofit for corrupted-but-present sensing without retraining the deployed detector.

发表机构

• University of Science, VNU-HCM（越南国立大学胡志明市理科大学）
• Michigan State University（密歇根州立大学）
• Center for Environmental Intelligence, VinUniversity（VinUniversity 环境智能中心）
• University of Arkansas（阿肯色大学）

机构由 AI 辅助整理，请以论文原文为准。