LOIC: Open-World Instance Classification in LiDAR Point Clouds

Wenbang Deng

doi:10.1109/tase.2026.3729369

LOIC: Open-World Instance Classification in LiDAR Point Clouds

2026-09-01

0

PRE

AI

DOI：10.1109/tase.2026.3729369

Original

Original request for help

Save

Abstract

En 中文

Open-world instance classification in LiDAR point clouds aims to discover and categorize instances of unknown classes unseen during training, without access to their ground-truth labels. The task is extremely challenging since only known-class labels are available for training. Existing LiDAR classification methods often depend on restrictive priors, including a predefined number of novel classes or additional annotations for unseen objects, which limits their applicability in continually changing environments. In this paper, we propose LOIC, an open-world instance classification framework for LiDAR point clouds with instance-level contrastive learning. LOIC first obtains candidate unknown instances and their voxel features through open-set semantic segmentation and ellipsoidal clustering. The introduced Classification Feature Extraction Head module (CFEH) then transforms point-wise features into discriminative cluster-based instance representations. The proposed Classification with Geometry-aware High-dimensional Features module (CGHF) further integrates orientation-robust geometric cues with the learned representations. Unknown instances are continuously assigned to discovered categories or initialized as novel categories according to their distances from dynamically maintained class prototypes, without requiring the unknown-class number in advance. We establish an evaluation benchmark on the nuScenes, SemanticKITTI, and in-house Campus datasets. Experiments show that LOIC outperforms state-of-the-art approaches adapted from the visual and feature-clustering domains, achieving improvements by 3.0, 4.8, and 7.9 percentage points in PQ, $\text {mIoU}_{\text {u}}$ , and $\text {mPre}_{\text {u}}$ , respectively. The source code of our method will be publicly available at https://github.com/nubot-nudt/LOIC Note to Practitioners—Autonomous robots frequently encounter objects whose categories are not represented in their training data. This paper presents an open-world instance classification framework for LiDAR point clouds, continuously discovering and classifying unseen instances of unknown categories. The proposed method can support safer navigation and more reliable environmental understanding by enabling robots to adopt object-specific strategies for unfamiliar obstacles. In addition, the framework may help to reduce manual annotation efforts through continual pseudo-label generation for newly discovered categories. The current implementation depends on the accuracy of upstream open-set segmentation and instance clustering. Future work will focus on improving robustness in dense and highly dynamic environments.

Keywords:

LiDAR

open world

instance classification

feature extraction

LOIC: Open-World Instance Classification in LiDAR Point Clouds

Abstract

Journal

Organization

Cited Papers