Learned Non-Maximum Suppression for 3D Object Detection
For 3D object detection practitioners, this provides a principled alternative to heuristic NMS that enhances detector reliability without modifying the base network.
This work introduces two learned filtering modules (D2D-Rescore and GossipNet3D) that replace heuristic NMS in LiDAR-based 3D object detection, improving mAP, NDS, and true positive quality, especially for small and infrequent classes, with minimal computational overhead.
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 .