用可学习的过滤机制替代传统NMS，提升3D目标检测精度。
Learned Non-Maximum Suppression for 3D Object Detection
- 基于检测间注意力和局部消息传递，实现智能去重
- 在nuScenes上提升小物体检测性能，mAP与NDS均优化
- 无需改动主干网络，适合部署于现有3D检测系统
LiDAR-based 3D目标检测中的后处理阶段至关重要，需对密集且重叠的候选框进行筛选以获得紧凑可靠的感知结果。本文提出两种可学习的过滤模块，取代传统的启发式非极大值抑制（NMS）。D2D-Rescore采用基于Transformer的检测到检测（D2D）注意力机制，GossipNet3D则将2D GossipNet思想拓展至3D，通过鸟瞰图中的局部消息传递实现。一种与nuScenes评估协议一致的度量感知匹配策略确保了训练与验证行为的一致性，提升了整体检测性能。两种方法在小类和罕见类上均显著优于CircleNMS，同时在保持极低计算开销的前提下，提升了平均精度（mAP）、nuScenes检测得分（NDS）及真正例质量。结果表明，可学习的检测级过滤可在不修改基础网络的情况下增强3D检测器的可靠性，提供了一种比启发式抑制更合理的新方案。代码已开源：https://github.com/rst-tu-dortmund/learned-3d-nms。
原文摘要 · Abstract (English)
Post-processing is a critical stage in LiDAR-based 3D object detection, where dense and overlapping proposals must be filtered for compact and reliable perception. This work introduces two learned filtering modules that replace heuristic non-maximum suppression (NMS) by leveraging relations among detections. D2D-Rescore employs transformer-based detection-to-detection (D2D) attention, while GossipNet3D adapts the 2D GossipNet concept to 3D through localized message passing in bird's-eye view. A metric-aware matching strategy aligned with the nuScenes evaluation protocol ensures consistent training and validation behavior, improving overall detection performance. Both approaches improve mean average precision (mAP), nuScenes detection score (NDS), and true positive quality compared to CircleNMS, particularly for small and infrequent classes, while adding minimal computational overhead. These results demonstrate that learned, detection-level filtering can enhance 3D detector reliability without modifying the base network, offering a principled alternative to heuristic suppression. Code is available at https://github.com/rst-tu-dortmund/learned-3d-nms .
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。