Research map / Point clouds / Segmentation and detection

BeyondMix: Leveraging Structural Priors and Long-Range Dependencies for Domain-Invariant LiDAR Segmentation

NeurIPS 2025 · Point clouds › Segmentation and detection · approach: LiDAR segmentation for driving

Read the paperCheck your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes BeyondMix, a mix-based unsupervised domain adaptation method for LiDAR semantic segmentation that exploits permutation invariance, local consistency, and geometric consistency priors along with long-range dependencies.

Problem

Unsupervised domain adaptation for LiDAR semantic segmentation across sensor/domain shifts, where existing mix-based methods fail to fully exploit the structural properties of point clouds.

Approach

Identifies three underexploited structural priors of LiDAR point clouds (permutation invariance, local consistency, geometric consistency) and builds a mix-based adaptation framework that leverages them together with long-range dependency modeling to learn domain-invariant features.

Experimental setup

Builds on: mix-based domain adaptation for LiDAR segmentation

Closest papers by meaning

• BeyondSparse: Facilitating Mamba to Enhance Cross-Domain 3D Semantic Segmentation in Adverse Weather (AAAI 2026 · similarity 0.91)
• Complete & Label: A Domain Adaptation Approach to Semantic Segmentation of LiDAR Point Clouds (CVPR 2021 · similarity 0.90)
• Less Is More: Reducing Task and Model Complexity for 3D Point Cloud Semantic Segmentation (CVPR 2023 · similarity 0.90)
• Construct to Associate: Cooperative Context Learning for Domain Adaptive Point Cloud Segmentation (CVPR 2024 · similarity 0.90)
• D^3CTTA: Domain-Dependent Decorrelation for Continual Test-Time Adaption of 3D LiDAR Segmentation (CVPR 2025 · similarity 0.89)
• TREND: Unsupervised 3D Representation Learning via Temporal Forecasting for LiDAR Perception (NeurIPS 2025 · similarity 0.87)