aiqbench analysis
SplatLabel advances 3D scene understanding by utilizing a 4D Gaussian representation to automate LiDAR segmentation and occupancy prediction without requiring pre-annotated bounding boxes. By integrating temporal manifolds for dynamic tracking and distilling probabilities from 2D foundation models, the method effectively resolves semantic ambiguities in complex environments. This approach represents a significant step forward in leveraging vision foundation models for robust autonomous driving perception tasks.
SplatLabel: Pseudo-Labelling through 4D Gaussian Splatting
arXiv:2609.29836v1 Announce Type: cross Abstract: While 2D Vision Foundation Models offer a pathway to automate 3D semantic pseudo-labelling, translating these priors into robust 3D representations typically requires complex heuristics or multi-model ensembles. We introduce SplatLabel, an automated pipeline that leverages a 4D Gaussian representation to extract LiDAR segmentation with predictive confidence, as well as semantic occupancy grids at arbitrary voxel resolutions. At its core, SplatLabel handles dynamic environments through an explicit temporal manifold that models the trajectories and lifespans of individual 3D primitives. This allows the system to accurately track moving actors and strictly define when objects appear and disappear, completely eliminating the need for pre-annotated 3D bounding boxes. To robustly support this dynamic tracking, the representation is grounded by structural and semantic priors: we guide scene geometry in unobserved regions by integrating 360-degree LiDAR via virtual depth maps, and rather than relying on domain-specific prompt engineering, we directly distill continuous soft probabilities from 2D models to inherently resolve semantic ambiguities over time and space. Finally, to accurately reflect the real-world trade-off between precision and recall, we reframe pseudo-label evaluation as a selective classification task using a generalized risk-recall metric. Experiments on SemanticKITTI demonstrate that SplatLabel consistently outperforms state-of-the-art baselines across multiple recall levels, establishing a highly robust framework for both 3D LiDAR segmentation and occupancy prediction.
Abstract:While 2D Vision Foundation Models offer a pathway to automate 3D semantic pseudo-labelling, translating these priors into robust 3D representations typically requires complex heuristics or multi-model ensembles. We introduce SplatLabel, an automated pipeline that leverages a 4D Gaussian representation to extract LiDAR segmentation with predictive confidence, as well as semantic occupancy grids at arbitrary voxel resolutions. At its core, SplatLabel handles dynamic environments through an explicit temporal manifold that models the trajectories and lifespans of individual 3D primitives. This allows the system to accurately track moving actors and strictly define when objects appear and disappear, completely eliminating the need for pre-annotated 3D bounding boxes. To robustly support this dynamic tracking, the representation is grounded by structural and semantic priors: we guide scene geometry in unobserved regions by integrating 360-degree LiDAR via virtual depth maps, and rather than relying on domain-specific prompt engineering, we directly distill continuous soft probabilities from 2D models to inherently resolve semantic ambiguities over time and space. Finally, to accurately reflect the real-world trade-off between precision and recall, we reframe pseudo-label evaluation as a selective classification task using a generalized risk-recall metric. Experiments on SemanticKITTI demonstrate that SplatLabel consistently outperforms state-of-the-art baselines across multiple recall levels, establishing a highly robust framework for both 3D LiDAR segmentation and occupancy prediction.
Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
arXiv:2609.29836 [cs.CV]
(or arXiv:2609.29836v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2609.29836 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 311 words
Related coverage
Open-Vocabulary BEV Segmentation with 3D-Aware Geometric Constraints
arXiv · Jul 31, 2026
Geometric Regularization for Long-Tailed Semi-Supervised Learning via Gaussian Feature Bridges
arXiv · Aug 24, 2026
The Role of Pseudo-labels in Self-training Linear Classifiers on High-dimensional Gaussian Mixture Data
arXiv · Jul 27, 2026
Pseudo-Label Augmentation for Affect Sensing in Small Collaborative Groups
arXiv · Sep 16, 2026