A direct feature-space approach for 3D LiDAR anomaly segmentation achieves competitive results on existing and new mixed real-synthetic datasets.
Spherical transformer for lidar-based 3d recognition
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.CV 2years
2026 2verdicts
UNVERDICTED 2representative citing papers
LEADER cuts position error by 24.1% on Oxford RobotCar and 73.9% on NCLT by combining a projection-based geometric encoder with a truncated relative reliability loss that down-weights unreliable points.
citing papers explorer
-
Learning to Identify Out-of-Distribution Objects for 3D LiDAR Anomaly Segmentation
A direct feature-space approach for 3D LiDAR anomaly segmentation achieves competitive results on existing and new mixed real-synthetic datasets.
-
LEADER: Learning Reliable Local-to-Global Correspondences for LiDAR Relocalization
LEADER cuts position error by 24.1% on Oxford RobotCar and 73.9% on NCLT by combining a projection-based geometric encoder with a truncated relative reliability loss that down-weights unreliable points.