Can we Trust Unreliable Voxels? Exploring 3D Semantic Occupancy Prediction under Label Noise

• URL: http://arxiv.org/abs/2603.06279v1
• Date: Fri, 06 Mar 2026 13:40:09 GMT
• Title: Can we Trust Unreliable Voxels? Exploring 3D Semantic Occupancy Prediction under Label Noise
• Abstract summary: 3D semantic occupancy prediction is a cornerstone of robotic perception.<n>Real-world voxel annotations are inherently corrupted by structural artifacts and dynamic trailing effects.<n>This raises a critical but underexplored question: can autonomous systems safely rely on such unreliable occupancy supervision?
• Score: 32.19984914850378
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: 3D semantic occupancy prediction is a cornerstone of robotic perception, yet real-world voxel annotations are inherently corrupted by structural artifacts and dynamic trailing effects. This raises a critical but underexplored question: can autonomous systems safely rely on such unreliable occupancy supervision? To systematically investigate this issue, we establish OccNL, the first benchmark dedicated to 3D occupancy under occupancy-asymmetric and dynamic trailing noise. Our analysis reveals a fundamental domain gap: state-of-the-art 2D label noise learning strategies collapse catastrophically in sparse 3D voxel spaces, exposing a critical vulnerability in existing paradigms. To address this challenge, we propose DPR-Occ, a principled label noise-robust framework that constructs reliable supervision through dual-source partial label reasoning. By synergizing temporal model memory with representation-level structural affinity, DPR-Occ dynamically expands and prunes candidate label sets to preserve true semantics while suppressing noise propagation. Extensive experiments on SemanticKITTI demonstrate that DPR-Occ prevents geometric and semantic collapse under extreme corruption. Notably, even at 90% label noise, our method achieves significant performance gains (up to 2.57% mIoU and 13.91% IoU) over existing label noise learning baselines adapted to the 3D occupancy prediction task. By bridging label noise learning and 3D perception, OccNL and DPR-Occ provide a reliable foundation for safety-critical robotic perception in dynamic environments. The benchmark and source code will be made publicly available at https://github.com/mylwx/OccNL.

Related papers

• LetOccVote: Learning Weakly Supervised 3D Occupancy through Consensus [20.730599565199935]
  LetOccVote is a weakly supervised framework for 3D occupancy learning.<n>It is trained solely with 2D pseudo-label supervision without requiring 3D occupancy annotations.<n>It achieves 53.27 IoU and 20.39 mIoU, establishing state-of-the-art performance among methods with 2D pseudo-label supervision.
  arXiv  Detail & Related papers  (2026-09-04T08:03:57Z)
• LESV: Language Embedded Sparse Voxel Fusion for Open-Vocabulary 3D Scene Understanding [9.377694035678948]
  We present a novel framework that leverages Sparse Voxel Rasterization (SVRaster) as a structured, disjoint geometry representation.<n>This enables a deterministic, confidence-aware feature registration process and suppresses the semantic bleeding artifact common in 3DGS.<n>Our approach achieves state-of-the-art performance on Open Vocabulary 3D Object Retrieval and Point Cloud Understanding benchmarks.
  arXiv  Detail & Related papers  (2026-04-01T20:48:06Z)
• Generalizing Abstention for Noise-Robust Learning in Medical Image Segmentation [2.597921446818458]
  abstention mechanism has proven effective in classification tasks by enhancing the capabilities of Cross Entropy.<n>We introduce a universal and modular abstention framework capable of enhancing the noise-robustness of a diverse range of loss functions.<n>Our framework improves upon prior work with two key components: an informed regularization term to guide abstention behaviour, and a more flexible power-law-based auto-tuning algorithm for the abstention penalty.
  arXiv  Detail & Related papers  (2026-01-20T14:57:56Z)
• Photorealistic Phantom Roads in Real Scenes: Disentangling 3D Hallucinations from Physical Geometry [17.571800971505787]
  This paper introduces the first end-to-end framework to probe, quantify, and tame this unquantified safety risk.<n>To probe, we present 3D-Mirage, the first benchmark of real-world illusions with precise planar-region annotations and context-restricted crops.<n>To quantify, we propose a Laplacian-based evaluation framework with two metrics: the Deviation Composite Score (DCS) for spurious non-planarity and the Confusion Composite Score (CCS) for contextual instability.
  arXiv  Detail & Related papers  (2025-12-17T13:14:37Z)
• Exploring Single Domain Generalization of LiDAR-based Semantic Segmentation under Imperfect Labels [28.96799571666756]
  We introduce the novel task Domain Generalization for LiDAR under Noisy Labels (DGLSS-NL)<n>We find that existing noisy-label learning approaches adapt poorly to LiDAR data.<n>We propose DuNe, a dual-view framework with strong and weak branches that enforce feature-level consistency and apply cross-entropy loss based on confidence-aware filtering of predictions.
  arXiv  Detail & Related papers  (2025-10-10T06:11:34Z)
• Out-of-Distribution Semantic Occupancy Prediction [26.178096447889605]
  We introduce Out-of-Distribution Semantic Occupancy Prediction, targeting OoD detection in 3D voxel space.<n>We introduce OccOoD, a novel framework integrating OoD detection into 3D semantic occupancy prediction, with Voxel-BEV Progressive Fusion (VBPF)<n> Experimental results demonstrate that OccOoD achieves state-of-the-art OoD detection with an AuROC of 67.34% and an AuPRCr of 29.21% within a 1.2m region, while maintaining competitive occupancy prediction performance.
  arXiv  Detail & Related papers  (2025-06-26T12:44:29Z)
• OccLE: Label-Efficient 3D Semantic Occupancy Prediction [68.60633561134571]
  OccLE is a Label-Efficient 3D Semantic Occupancy Prediction.<n>It takes images and LiDAR as inputs and maintains high performance with limited voxel annotations.<n> Experiments show that OccLE achieves competitive performance with only 10% of voxel annotations.
  arXiv  Detail & Related papers  (2025-05-27T01:41:28Z)
• Combating Bilateral Edge Noise for Robust Link Prediction [56.43882298843564]
  We propose an information-theory-guided principle, Robust Graph Information Bottleneck (RGIB), to extract reliable supervision signals and avoid representation collapse. Two instantiations, RGIB-SSL and RGIB-REP, are explored to leverage the merits of different methodologies. Experiments on six datasets and three GNNs with diverse noisy scenarios verify the effectiveness of our RGIB instantiations.
  arXiv  Detail & Related papers  (2023-11-02T12:47:49Z)
• Dynamics-Aware Loss for Learning with Label Noise [73.75129479936302]
  Label noise poses a serious threat to deep neural networks (DNNs) We propose a dynamics-aware loss (DAL) to solve this problem. Both the detailed theoretical analyses and extensive experimental results demonstrate the superiority of our method.
  arXiv  Detail & Related papers  (2023-03-21T03:05:21Z)
• Semantic Scene Completion with Cleaner Self [93.99441599791275]
  Semantic Scene Completion (SSC) transforms an image of single-view depth and/or RGB 2D pixels into 3D voxels, each of whose semantic labels are predicted. SSC is a well-known ill-posed problem as the prediction model has to "imagine" what is behind the visible surface, which is usually represented by Truncated Signed Distance Function (TSDF) We use the ground-truth 3D voxels to generate a perfect visible surface, called TSDF-CAD, and then train a "cleaner" SSC model. As the model is noise-free, it is expected to
  arXiv  Detail & Related papers  (2023-03-17T13:50:18Z)
• UpCycling: Semi-supervised 3D Object Detection without Sharing Raw-level Unlabeled Scenes [7.32610370107512]
  UpCycling is a novel SSL framework for 3D object detection with zero additional raw-level point cloud. We introduce hybrid pseudo labels, feature-level Ground Truth sampling (F-GT) and Rotation (F-RoT) UpCycling significantly outperforms the state-of-the-art SSL methods that utilize raw-point scenes.
  arXiv  Detail & Related papers  (2022-11-22T02:04:09Z)
• GLENet: Boosting 3D Object Detectors with Generative Label Uncertainty Estimation [70.75100533512021]
  In this paper, we formulate the label uncertainty problem as the diversity of potentially plausible bounding boxes of objects. We propose GLENet, a generative framework adapted from conditional variational autoencoders, to model the one-to-many relationship between a typical 3D object and its potential ground-truth bounding boxes with latent variables. The label uncertainty generated by GLENet is a plug-and-play module and can be conveniently integrated into existing deep 3D detectors.
  arXiv  Detail & Related papers  (2022-07-06T06:26:17Z)
• Towards Adversarially Robust Deep Image Denoising [199.2458715635285]
  This work systematically investigates the adversarial robustness of deep image denoisers (DIDs) We propose a novel adversarial attack, namely Observation-based Zero-mean Attack (sc ObsAtk) to craft adversarial zero-mean perturbations on given noisy images. To robustify DIDs, we propose hybrid adversarial training (sc HAT) that jointly trains DIDs with adversarial and non-adversarial noisy data.
  arXiv  Detail & Related papers  (2022-01-12T10:23:14Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.