Article

Cut a Wall, See a Wall: Identity Substitution in Occluded Point-Cloud Perception for Construction Robots

The paper finds that when entire regions of a scene are occluded at test time, point-cloud segmentation networks relabel the hidden elements into a single sink class; under fixed layouts this class absorbs up to 70% of test points, and baseline mIoU drops to 0.15-0.27. The authors append 14 hand-computed geometric features per point (GPE), restoring mIoU to 0.54-0.92 on PointNet++ and a PTv3-style transformer and removing 64-100% of whole-object losses; on real S3DIS scans, whole-wall losses drop from 10 to 0.

Part of the event: Occluded point cloud segmentation networks mislabel hidden components as the same category

This page provides a summary and source link only.